System and method for predicting and mitigating cybersecurity system misconfigurations

Abstract

A computerized method for reconfiguring one or more malware detection systems each performing cybersecurity analyses on incoming data is described. The method involves receiving meta-information including metrics associated with a malware detection system. Based on the meta-information, a determination is made whether the malware detection system is operating at an optimal performance level. If not, results produced by conducting behavior analyses predicting operability of the malware detection system are determined and the results are provided as feedback to the malware detection system to update one or more configuration parameter values thereof.

Description

FIELD

Embodiments of the disclosure relate to cyber security. More particularly, one embodiment of the disclosure relates to a system and method for predicting the current performance level of a malware detection system and altering its configuration based on the predicted performance level.

GENERAL BACKGROUND

Network devices provide useful and necessary services that assist individuals in business and in their everyday lives. In recent years, a growing number of cyberattacks are being conducted on all types of network devices, especially network devices deployed at an enterprise (e.g., private or publicly-traded company, a governmental agency, etc.). In some cases, these cyberattacks are orchestrated in an attempt to gain access to content stored on one or more of these enterprise-based network devices. Such access is for illicit (i.e., unauthorized) purposes, such as spying or other malicious or nefarious activities. For protection, many enterprises deploy cybersecurity systems, such as on-premises malware detection systems that monitor and analyze content propagating over a local network in efforts to detect a cyberattack.

Typically, on-premises malware detection systems are installed in accordance with configurations that are either factory set or user-configurable, e.g., per specifications of installation guides provided by the manufacturers. Typically, these malware detection systems are initially configured to operate efficiently in accordance with network traffic patterns generally prevailing at the time of installation. Sometimes, the malware detection systems may not be properly configured by customers (users). Moreover, as network traffic patterns are dynamic and the threat landscape confronting customers may differ and may even change over time, in some situations, the malware detection systems' configurations should be tuned upon installation and re-tuned from time to time after installation for optimal its effectiveness.

In extreme situations, the malware detection systems may be significantly misconfigured, resulting in under-utilization of their detection capabilities or over-utilization of their detection capacity. This may reduce operational efficiencies or efficacies, or both, below that achievable by properly configured malware detection systems, and, in worse case scenarios, may result in inadequately analyzed network traffic and unnecessarily increased risk of a successful cyberattack on an enterprise. However, this activity of reconfiguring or tuning of the malware detection system is rarely performed due to both increased costs for the customer and a reduced availability of skilled technicians to perform such services.

BRIEF DESCRIPTION OF THE DRAWINGS

Embodiments of the disclosure are illustrated by way of example and not by way of limitation in the figures of the accompanying drawings, in which like references indicate similar elements and in which:

FIG. 1 is an exemplary embodiment of an architecture of a cybersecurity protection service.

FIG. 2 is an exemplary embodiment of a malware detection system of FIG. 1.

FIG. 3 is an exemplary embodiment of the system configuration optimization engine of FIG. 1.

FIG. 4 is a first exemplary embodiment of an operation flow between a cybersecurity system and the system configuration optimization engine of FIG. 1.

FIG. 5 is an exemplary embodiment of an interactive dashboard of a first interface display screen produced by the monitoring and reporting service of FIG. 4.

FIG. 6 is an exemplary embodiment of an interactive dashboard of a second interface display screen to provide greater granularity in the analytic results illustrated by the first display screen of FIG. 5.

FIG. 7 is an exemplary embodiment of the logical operations performed during communications between a malware detection system deployed within a cybersecurity system and the system configuration optimization engine.

FIG. 8 is an exemplary embodiment of a flowchart illustrating operations by the system configuration optimization engine of FIG. 1.

FIG. 9 is a second exemplary embodiment of an operation flow between a cybersecurity system and the system configuration optimization engine of FIG. 1.

DETAILED DESCRIPTION

I. Overview

In general, embodiments of the disclosure describe a system configuration optimization engine that is configured to (i) receive meta-information including different metrics associated with one or more malware detection systems (situated on-premises or as part of a cloud service), (ii) determine whether each of the malware detection system(s) is operating at an optimal performance level, and (iii) generate results provided as feedback to update one or more configuration parameter values for a particular malware detection system that is operating at a non-optimal performance level. Each configuration parameter includes information that partially controls the operating state of a resource (e.g., hardware, software or firmware) deployed within a network device (e.g., malware detection system). Examples of configuration parameters may be directed to hardware characteristics (e.g., number of active processor cores, memory capacity, utilization levels, etc.), operational settings such as virtual machine (VM) characteristics (e.g., number of active VMs, VM utilization, etc.), kernel related optimization (e.g. enable/disable kernel filters according to operating system performance, etc.) software characteristics (e.g., number of active processes, applications utilized by the active processes, etc.) or the like.

As described below, the system configuration optimization engine is remotely located from and communicatively coupled to one or more cybersecurity systems, which may be associated with different customers. A cybersecurity system includes one or more malware detection systems, each configured to provide meta-information to the system configuration optimization engine. The meta-information may include (a) a first metric being statistics associated with certain configuration parameters of the malware detection system, and/or (b) a second metric being events monitored during operations of the malware detection system (e.g., data retrieved in response to a certain event such as a system crash, etc.).

Based on the received meta-information, the system configuration optimization engine assigns a performance level for each malware detection system of a cybersecurity system. For illustrative purposes, the assigned performance level may be one of a plurality of performance levels, either (i) an optimal performance level or (ii) a non-optimal performance level. The non-optimal performance level may include multiple levels of granularity, such as an over-utilized performance level and an under-utilized performance level. The “optimal performance level” refers to a preferred operating state for a network device performing cybersecurity analyses, such as performed by a malware detection system for example, which may be measured by certain metrics, such as hardware utilization statistics, virtual machine utilization statistics, and/or software utilization statistics. A “non-optimal performance level” (e.g., over-utilized or under-utilized) identifies the malware detection system is operating outside of its desired operating configuration.

More specifically, as described herein, the system configuration optimization engine analyzes the meta-information provided by each malware detection system and, based on such analysis, assigns a performance level to that malware detection system. The meta-information may be provided to the system configuration optimization engine on a periodic basis and/or an aperiodic basis in response to a certain event (e.g., system crash, system operability exceeds or falls below a prescribed threshold, request initiated by a network administrator or cybersecurity system manufacturer, etc.).

As described above, the “optimal” performance level refers to a preferred operating state for a network device preforming cybersecurity analyses. This preferred operating state may be represented through a collection of system metrics, and thus, the performance level for a malware detection system may be determined through the collective analysis of configuration parameter values directed to these system metrics, which are provided as part of the meta-information. As part of this collective analysis, the system configuration optimization engine determines the degree of correlation between the received metrics in the meta-information and desired metrics of the malware detection system (referred to as a “baseline configuration”) that are gathered using experiential knowledge of operational meta-information of known misconfigured malware detection systems and/or known optimal malware detection systems.

According to one embodiment of the disclosure, a baseline configuration may include certain hardware utilization threshold (or range), VM utilization threshold (or range), and/or software utilization threshold (or range) representing a desired operating configuration for a malware detection system with a certain hardware profile. Hence, the system configuration optimization engine may include a plurality of baseline configurations each associated with a different hardware profile (e.g., number of processors, memory size, etc.). When analyzing the performance level for a malware detection system, a baseline configuration may be selected based on the hardware profile for that malware detection system, along with other factors (e.g., threat landscape confronting the customer (for instance, as indicated by the industry protected by the malware detection system), subscription or customer type, etc.). For instance, given the same hardware profile, a malware detection system deployed for a customer in a high-risk industry (e.g., governmental defense agency, utility, etc.) may be assigned a more stringent baseline configuration (e.g., less range tolerance, different weighting scheme targeted to ensure higher average performance levels, etc.) than a baseline configuration reserved for malware detection systems deployed in lower-risk industries (e.g., textiles, etc.). As a result, subscription levels for malware detection systems deployed for high-risk industry customers may be more costly given a likely increased frequency of re-configuration of the malware detection system for placement into an optimal performance level.

According to one embodiment of the disclosure, the performance level for the malware detection system is determined by conducting one or more arithmetic or logical operations on performance level determinations conducted for a plurality of metrics provided as part of the meta-information. More specifically, each statistic of a configuration parameter may be assigned a weighting depending on its importance in representing the health of the malware detection system. For instance, a first statistic directed to processor utilization may be assigned a larger weighting (i.e., assigned a higher importance) than a second metric directed to the number of virtual machines (VMs) currently active. Hence, the performance level for the malware detection system is based on a collection of weighted, performance level determinations (e.g., over-utilized, optimal, under-utilized) based on the statistics provided as part of the meta-information.

Responsive to the meta-information, the system configuration optimization engine may be configured to return information (referred to as a “system health report”) to the malware detection system supplying the meta-information. For one embodiment, the system health report includes (i) an identifier of the malware detection system supplying the analyzed meta-information; (ii) a determined performance level for the malware detection system; (iii) the performance level determinations for some or all of the plurality of configuration parameters; and/or (iv) one or more modified configuration parameter values that are used by the malware detection system to adjust its configuration to better remain in or return to an optimal performance level.

The system configuration optimization engine may be configured, prior to selection and passing of one or more modified configuration parameter values to the malware detection system, to select the modified configuration parameter values by at least comparing the received meta-information to predefined operational bounds (e.g., a blacklist including one or more statistics associated with configuration parameters for devices with the same hardware profile operating at non-optimal performance levels (e.g., misconfigured systems, etc.), and/or whitelist including statistics of configuration parameters for devices with the same hardware profile operating at optimal performance levels). Additionally, the system configuration optimization engine may provide the modified configuration parameter values to a monitoring and reporting service, which may issue an alert to a customer of the malware detection systems upon determining that one or more of the malware detection systems is operating at a non-optimal performance level.

According to one embodiment of the disclosure, the configuration of a malware detection system operating at a non-optimal performance level may be updated automatically without customer approval. Alternatively, before configuration parameter value(s) for the particular malware detection system are updated, approval from the network administrator is needed. Herein, administrator approval may be secured by a network device implemented with the system configuration optimization engine prior to providing the results (i.e., one or more modified configuration parameter value(s) as feedback to the particular malware detection system.

As an illustrative example, the network device may send an alert message to the administrator (e.g., text, email, notice to access a dashboard, etc.), where the alert message requires an action by the administrator before the modified configuration parameter values are sent. The action may include, but is not limited or restricted to any reply mechanism such as selection of a radio button, selection of a display element (or entry of information) on the dashboard, or the like. As another illustrative example, the network device may send the modified configuration parameter values to the particular malware detection system and provide the results as input to a monitor/reporting service, which generates a display that is accessible by at least an authorized administrator and illustrates performance levels of the malware detection systems utilized by a customer. Any of these types of reply mechanisms allows the network administrator to interact with the dashboard to authorize the configuration parameter update.

II. Terminology

In the following description, certain terminology is used to describe various features of the invention. For example, each of the terms “logic,” “engine,” and “component” may be representative of hardware, firmware or software that is configured to perform one or more functions. As hardware, the term logic (or engine or component) may include circuitry having data processing and/or storage functionality. Examples of such circuitry may include, but are not limited or restricted to a hardware processor (e.g., microprocessor, one or more processor cores, a digital signal processor, a programmable gate array, a microcontroller, an application specific integrated circuit “ASIC”, etc.), a semiconductor memory, or combinatorial elements.

Additionally, or in the alternative, the logic (or engine or component) may include software such as one or more processes, one or more instances, Application Programming Interface(s) (API), subroutine(s), function(s), applet(s), servlet(s), routine(s), source code, object code, shared library/dynamic link library (dll), or even one or more instructions. This software may be stored in any type of a suitable non-transitory storage medium, or transitory storage medium (e.g., electrical, optical, acoustical or other form of propagated signals such as carrier waves, infrared signals, or digital signals). Examples of a non-transitory storage medium may include, but are not limited or restricted to a programmable circuit; non-persistent storage such as volatile memory (e.g., any type of random access memory “RAM”); or persistent storage such as non-volatile memory (e.g., read-only memory “ROM”, power-backed RAM, flash memory, phase-change memory, etc.), a solid-state drive, hard disk drive, an optical disc drive, or a portable memory device. As firmware, the logic (or component) may be stored in persistent storage.

Herein, a “message” generally refers to related data that is received, transmitted, or exchanged over a communication session. The message may include one or more packets, where a “packet” broadly refers to a series of bits or bytes having a prescribed format. Alternatively, the data may include a collection of data that may take the form of an individual or a number of packets carrying related payloads, e.g., a single webpage received over a network.

The term “object” generally relates to content (or information for accessing such content) having a logical structure or organization that enables the object to be classified for purposes of malware analysis. The content may include an executable (e.g., an application, program, code segment, a script, dynamic link library “dll” or any file in a format that can be directly executed by a computer such as a file with an “.exe” extension, etc.), a non-executable (e.g., a file; any document such as a Portable Document Format “PDF” document; a word processing document such as Word® document; an electronic mail “email” message, web page, etc.), or simply a collection of related data (e.g., packets).

The term “computerized” generally represents that any corresponding operations are conducted by hardware in combination with software and/or firmware. The term “data store” generally refers to a data storage device such as the non-transitory storage medium described above, which provides non-persistent or persistent storage for information (e.g., data, meta-information, etc.).

According to one embodiment of the disclosure, the term “malware” may be broadly construed as any code, communication or activity that initiates or furthers a cyberattack. Malware may prompt or cause unauthorized, anomalous, unintended and/or unwanted behaviors or operations constituting a security compromise of information infrastructure. For instance, malware may correspond to a type of malicious computer code that, as an illustrative example, executes an exploit to take advantage of a vulnerability in a network, network device or software, to gain unauthorized access, harm or co-opt operations of the network, the network device of the software or to misappropriate, modify or delete data. Alternatively, as another illustrative example, malware may correspond to information (e.g., executable code, script(s), data, command(s), etc.) that is designed to cause a network device to experience anomalous (unexpected or undesirable) behaviors. The anomalous behaviors may include a communication-based anomaly or an execution-based anomaly, which, for example, could (1) alter the functionality of a network device executing application software in an atypical manner; (2) alter the functionality of the network device executing that application software without any malicious intent; and/or (3) provide unwanted functionality which may be generally acceptable in another context.

The term “network device” may be construed as any electronic computing system with the capability of processing data and connecting to a network. The network may be a public network such as the Internet and/or a local (private) network such as an enterprise network, a wireless local area network (WLAN), a local area network (LAN), a wide area network (WAN), or the like. Examples of a network device may include, but are not limited or restricted to an endpoint (e.g., a laptop, a mobile phone, a tablet, a computer, a video console, a copier, etc.), a network appliance, a server, a router or other intermediary communication device, a firewall, etc.

The term “transmission medium” may be construed as a physical or logical communication path between two or more network devices or between components within a network device. For instance, as a physical communication path, wired and/or wireless interconnects in the form of electrical wiring, optical fiber, cable, bus trace, or a wireless channel using radio frequency (RF) or infrared (IR), may be used. A logical communication path may simply represent a communication path between two or more network devices or between components within a network device such as one or more Application Programming Interfaces (APIs).

Finally, the terms “or” and “and/or” as used herein are to be interpreted as inclusive or meaning any one or any combination. Therefore, “A, B or C” or “A, B and/or C” mean “any of the following: A; B; C; A and B; A and C; B and C; A, B and C.” An exception to this definition will occur only when a combination of elements, functions, steps or acts are in some way inherently mutually exclusive.

As this invention is susceptible to embodiments of many different forms, it is intended that the present disclosure is to be considered as an example of the principles of the invention and not intended to limit the invention to the specific embodiments shown and described.

III. General Architecture

Referring to FIG. 1, an exemplary block diagram of an embodiment of an architecture of a cybersecurity protection service 100 is shown. Herein, the cybersecurity protection service 100 features one or more cybersecurity systems 110₁-110_N (N≥1) communicatively coupled to a system configuration optimization engine 150. Deployed for detecting and protecting a local network of a customer against cyberattacks, each of the cybersecurity systems 110₁-110_N (e.g., cybersecurity system 110₁) includes one or more malware detection systems 120₁-120_M (M≥1). As shown, the malware detection systems 120₁-120_M may be deployed on-premises (in the local network) to detect and analyze incoming objects propagating into or through the local network for malware. Alternatively, the malware detection systems 120₁-120_M may be deployed as a cloud-based solution in which the incoming objects (or a representation thereof) are captured at the local network and provided to at least one of the cloud-based malware detection systems 120₁-120_M.

Herein, each of the malware detection system 120₁-120_M may be configured to perform a two-phase approach for detecting malware contained in network traffic. This two-phase approach includes a static phase and a dynamic phase. During the static phase, an initial analysis of the characteristics of an object is conducted, without execution or processing of the object, to determine whether the object is “malicious” (malware) or “benign” (not malware). Where the object is “suspicious” (e.g., inconclusive if malicious or benign), a further (dynamic) analysis of the object may be conducted. During the dynamic phase, the object is executed within one or more virtual machines. A virtual machine (VM) executes an incoming object and the behaviors of these objects during execution (or the VM) are monitored. Each behavior may also referred to as an “event.” In some embodiments, one or more malware detection system 120₁-120_M (e.g., malware detection system 120₁) may be deployed as a network device, which is communicatively coupled to receive and analyze objects within network traffic. As a network device, the malware detection system 120₁ includes logic being physical components that analyze objects for malware. Alternatively, the malware detection system 120₁ may be deployed as a virtual device, namely a software (daemon) agent to detect cyberattacks that is operating (in the foreground or background) within a network device (e.g., an endpoint). An example of the two-phase malware detection system is described in U.S. Pat. No. 9,311,479, entitled, “Correlation and Consolidation of Analytic Data For Holistic View of A Malware Attack,” and U.S. Pat. No. 9,483,644 entitled “Methods for Detecting File Altering Malware in VM based Analysis,” the entire contents of both of which are hereby incorporated by reference.

As shown in FIG. 1, each malware detection system 120₁, . . . , or 120_M is configured to transmit meta-information 130 to the system configuration optimization engine 150. Such transmission may be initiated periodically upon detection of a timeout condition (e.g., prescribed amount of time has elapsed, prescribed count value achieved, certain period of time detected, etc.) or aperiodically upon detection of a predetermined event (e.g., a system crash, completion of analyses of a prescribed number of objects, etc.). For instance, from the malware detection system 120₁, the meta-information 130 may include statistics 135 associated with one or more configuration parameters that identify a current operating state of the malware detection system 120₁. The statistics 135 may include, but are not limited or restricted to, measured values associated with certain configuration parameters, such as hardware utilization statistics, virtual machine utilization statistics, and/or software utilization statistics, as described above. Additionally, or in the alternative, the meta-information may include events as described herein.

According to one embodiment of the disclosure, for any deployment of a malware detection system, certain component(s) within the malware detection system 120₁ periodically or aperiodically determine the current operating state of the malware detection system. As an illustrative example, as shown in FIGS. 1-2, system configuration analysis logic 265 may be configured, in response to a timeout or predetermined event detected by monitoring logic 260, to ascertain statistics associated with certain configuration parameters. The statistics may be collected via the OS. The configuration parameter values may be temporarily stored prior to transmission to the system configuration optimization engine 150.

Referring back to FIG. 1, the system configuration optimization engine 150 may be implemented within a dedicated network device, which is located remotely from the cybersecurity systems 110₁-110_N. As an illustrative embodiment, the system configuration optimization engine 150 may be deployed as logic being part of public cloud computing or private cloud computing (e.g., private cloud, a virtual private cloud or a hybrid cloud). When operating as part of public cloud computing, the system configuration optimization engine 150 is accessible by each of the cybersecurity systems 110₁-110_N via a public network, as public cloud computing supports a multi-tenant environment. In contrast, when operating as part of private cloud computing, the system configuration optimization engine 150 is accessible by a single cybersecurity system (e.g., cybersecurity system 110₁) where each of the cybersecurity systems 110₁-110_N is associated with a different customer as private cloud computing supports a single-tenant environment.

Upon receipt of the meta-information 130 from the cybersecurity system 110₁, the system configuration optimization engine 150 analyzes the meta-information 130 to determine whether the malware detection system 120₁ is operating at an optimal performance level. Such analysis may involve an evaluation of the meta-information 130 against a predictive model based on heuristic information including configuration parameter values of a known body of past configurations (e.g., blacklist and/or whitelist and/or hardware profile) that are associated with non-optimal and/or optimal performance levels. By identifying the misconfigurations, the system configuration optimization engine 150 may recommend modifications to the malware detection system configuration. These modifications are alerted to the customer and/or directly provided to the malware detection system(s).

According to one embodiment of the disclosure, the determination whether the malware detection system 120₁ is operating at an optimal performance level, namely a preferred operating state for malware detection analyses as measured by metrics such as system parameter values and/or detection parameter values. Collectively, these configuration parameter values identify a health of the malware detection system 120₁ or the cybersecurity system 110₁ including at least the malware detection system 120₁. Herein, the system parameters may be directed to features that influence operability of the malware detection system 120₁, such as hardware utilization statistics (e.g., processor utilization, amount or degree of available memory for storage, etc.), virtual machine utilization statistics (e.g., virtual machine “VM” utilization or the number of VMs activated, etc.) and/or software utilization statistics (e.g., what processes are running, statistics associated with the processes, queue length, etc.). In contrast, the detection parameters may be directed to features associated with a malware detection analysis being conducted (e.g., type of analyses, duration of analysis per object, classification threshold being used to determine performance level, etc.

As an example, as described above, when adjusting the detection parameter values, the type of analysis may be varied (dynamic, emulation, types of static analysis, etc.) or the thresholds that determine suspiciousness (requiring further analysis e.g., dynamic analysis) may be varied in order to control the number of objects that are subjected to further analysis. For under-utilization, by changing the threshold to a lower level, we may subject more objects to deeper analysis (e.g., dynamic) which increases consumption of available system resources and increases the rate of object analysis. The additional objects subject to further analysis caused by a reduced threshold may have a lower probability (based on preliminary analysis only) of being malicious. However, such analysis may reduce the number of false negatives. For over-utilization, by raising the threshold, fewer objects may be subject to further (dynamic) analysis. As the threshold is related to likelihood of maliciousness, adjustment may be slow to ensure that there is no appreciable increase the risk of false negatives. The availability of such adjustments may be related to the prevailing threat landscape for the particular customer protected by the system, or its industry,

In response to determining that the malware detection system 120₁ is operating at a non-optimal performance level representing that the malware detection system 120₁ is operating outside of its desired operating configuration (e.g., under-utilized or over-utilized), the system configuration optimization engine 150 may be configured to recommend configuration modifications for the malware detection system 120₁ and return a configuration modification message 170. The configuration modification message 170 may include one or more modified configuration parameter values 175, which may be a different values than the received configuration parameter value 135. Upon modification of the configuration of the malware detection system 120₁ in accordance with the modified configuration parameter values 175, either automatically or upon approval by an administrator as described above before the automated solution is given effect (e.g., agreement as to the modifications, accept increased charges if applicable, etc.), the functionality of the malware detection system 120₁ is modified in efforts to return to an optimal performance level.

As an illustrative example, the configuration modification message 170 may include a modified configuration parameter value 175 signifying a change in the number of active virtual machines currently being utilized by the malware detection system 120₁, as represented by the statistics 135 (e.g., inclusive of statistics associated with the number of active virtual machines) provided within the meta-information 130. Herein, according to one embodiment, the malware detection system 120₁, upon receipt of the modified configuration parameter value 175 (with administrator approval if needed), may decrease the number of active virtual machines deployed, provided the malware detection system 120₁ is determined by the system configuration optimization engine 150 to be operating at an “over-utilized” VM utilization level. Herein, the “over-utilized” VM utilization level may be determined by any configuration parameter value or combination of configuration parameter values indicating that available resources at the malware detection system 120₁ are incapable of supporting the current performance level (e.g., the number of VMs running concurrently, number of objects queued and awaiting VM analysis, etc.), and the modified configuration parameter value 175 temporarily reducing the performance level of the malware detection system 120₁. Alternatively, according to another embodiment, upon receipt of the modified configuration parameter value 175 (with administrator approval if needed), the malware detection system 120₁ may increase in number of active virtual machines from the number of active virtual machines represented by the statistics 135, provided the malware detection system 120₁ is operating at an “under-utilized” VM utilization level where resources at the malware detection system 120₁ are available to support a higher performance level (e.g., more VMs, etc.).

Referring to FIG. 2, an illustrative embodiment of a malware detection system (e.g., malware detection system 120₁) is shown. Herein, the malware detection system 120₁ features a plurality of components, including a processor 200, a network interface 210, a memory 220, and an optional administrative interface 230, which are communicatively coupled together via a transmission medium 240. As shown, when deployed as a network appliance, the components are at least partially encased in a housing 250 made entirely or partially of a rigid material (e.g., hardened plastic, metal, glass, composite, or any combination thereof). The housing 250 protects these components from environmental conditions. As a virtual device, however, the malware detection system 120₁ includes some or all of the functionality provided by the logic within the memory 220.

The processor 200 is a multi-purpose, programmable component that accepts digital data as input, processes the input data according to stored instructions, and provides results as output. One example of a processor may include an Intel® central processing unit (CPU) with an x86 instruction set architecture. Alternatively, the processor 200 may include another type of CPU, a digital signal processor (DSP), an Application Specific Integrated Circuit (ASIC), a field-programmable gate array (FPGA), or the like.

As shown in FIG. 2, the processor 200 is communicatively coupled to the memory 220 via the transmission medium 240. According to one embodiment of the disclosure, the memory 220 is adapt to store (i) event/timeout monitoring logic 260, (ii) system configuration analysis logic 265, (iii) optional timestamp generation logic 270, (iv) meta-information storage logic 275, (v) configuration readjustment logic 280 and corresponding data store 285, and (vi) malware detection logic 290.

The configuration analysis logic 265, in response to a particular event or timeout detected by the monitoring logic 260, obtains meta-information (e.g., one or more configuration parameter values 135) associated with the current operating state of the malware detection system 120₁. As shown, the configuration analysis logic 265 includes an operating system (OS) statistics module 266 to collect hardware utilization statistics from the OS (e.g., processor utilization, amount or degree of available memory for storage, etc.); VM statistics module 267 to collect VM utilization statistics (e.g., VM utilization or the number of VMs activated, etc.); and/or application statistics module 268 to collect software utilization statistics (e.g., what processes are running, statistics associated with the processes, etc.).

Thereafter, the configuration analysis logic 265 may temporarily store statistics associated with certain configuration parameter(s) 135 (hereinafter, “statistics”) within the meta-information storage logic 275. As an optional operation, the timestamp generation logic 270 may generate a timestamp (not shown) that is applied to each value of the statistic 135 prior to storage with the meta-information storage logic 275. The statistics 135 may include the processor utilization level, the amount of hard disk space available, number of active virtual machines, the number of processes currently running, or the like. The current configuration parameter values 135, stored in the meta-information storage logic 275, are subsequently accessed from the meta-information storage logic 275 for transmission to the system configuration optimization engine 150 of FIG. 1.

The configuration readjustment logic 280 is adapted to receive the configuration information 170 (i.e., modified configuration parameter values 175) from the system configuration optimization engine 150 of FIG. 1 for storage with the data store 285. The configuration readjustment logic 280 (e.g., a script running on the malware detection system 120₁) is configured to change the operating state of the malware detection system 120₁ by altering certain configuration parameter values with the values included in the modified configuration parameter values 175. The adjustment of the current configuration parameter values, represented by the statistics 135, may occur upon receipt of the modified configuration parameter values 175 or after the configuration readjustment logic 280 initiates a message to an administrator to approve alteration of the operating state of the malware detection system 120₁ with the modified configuration parameter values 175. Alternatively, the current configuration parameter values may be adjusted in response to a change in operating state by the malware detection system 120₁ such as initialization of a new process or a time in which the processor utilization falls below a first predetermined value or exceeds a second predetermined value different than the first predetermined value.

The administrative interface 230 is a portal that allows an administrator, after credential exchange and authentication, to access and update logic stored within the memory 220 of the malware detection system 120₁. For instance, the administrative interface 230 may include authentication logic (not shown) to authenticate an administrator requesting access to stored logic within the malware detection system 120₁. Upon authentication, the administrator is able to modify (i) the triggering events or timeout parameters within the event/timeout monitoring logic 260, or (ii) code of the system configuration analysis logic 265, configuration readjustment logic 280, and/or malware detection logic 290 (e.g., code associated with static analysis of an object or the behavioral analysis of the object in efforts to detect a presence of malware within the object or its association with a cyberattack), or (iii) operability of the malware detection system 120₁ (e.g., hardware changes, operational setting changes or software changes as described below).

Referring now to FIG. 3, an exemplary embodiment of the system configuration optimization engine 150 of FIG. 1 is shown. The system configuration optimization engine 150 features one or more processors 300, a network interface 310, a memory 320, and an optional system administrative interface 330 which allows an administrator to directly access to data within the system configuration optimization engine 150 (e.g., adjust the code associated with any of the components set forth in the memory 320 when the administrator is authorized to perform such actions).

As shown, the memory 320 comprises a parser 350, training data storage 360, machine learning (ML) modeling logic 365, ML training model 370, ML predictive model 375, and system health reporting logic 380. Herein, the parser 350 is configured to parse both structured and non-structured data, which is provided as meta-information 130 from a malware detection system (e.g., malware detection system 120₁). More specifically, the parser 350 features a plurality of sub-parsers 355, including a first sub-parser 356 and a second sub-parser 357. The first sub-parser 356 is configured to parse structured data to recover meta-information including the values associated with one or more configuration parameters positioned at specific locations within the structured data (hereinafter, “recovered configuration parameter values”). The recovered configuration parameter values may be analyzed by the ML modeling logic 365 in accordance with the ML predictive model 375. The second sub-parser 357 is configured to parse unstructured data (e.g., line in a text file) for relevant information, including information associated with an event. For instance, the second sub-parser 357 may conduct a search for one or more keywords (e.g., “kernel crash” keyword, etc.) and extract information subsequent to the keywords (e.g., information identifying a nature and/or reason for the crash).

The training data storage 360 is a data store that is adapted to temporarily store sets of labeled training set 362 and/or unlabeled training set 363 (referred to as “training data” 364) for use by the machine learning modeling logic 365 in “training” the ML training model 370 to produce the ML predictive model 375. The training data storage 360 may include data from the cybersecurity systems 110₁-110_N as well as third party sources. Herein, the training data 364 include normalized, heuristic data pertaining to a plurality of configuration parameters directed to operability of a malware detection system, where some of the heuristic data may be directed to the same configuration parameter associated with a different hardware profile. The heuristic data may include a normalized value for a specific configuration parameter over a prescribed time period, as measured for a network device with a specific hardware profile over a prescribed time period. Alternatively, the heuristic data may include a prescribed number of sampled values. A “hardware profile” is a specific representation of a network device having certain functionality, such as a number of processing elements (e.g., processors, processor cores, etc.), certain memory storage capacity, certain VM capacity, manufacturer/model name of the network device, device identification number, or the like.

For different hardware profiles, each training data set 364 may correspond to a different configuration parameter, a different combination of configuration parameters, and/or different configuration parameter values or weighting used by different classifications. Stated differently, the ML modeling logic 365 uses the training data sets 364 to establish baselines in classifying incoming meta-information 130 (using the received configuration parameter values), and these baselines may vary between hardware profiles. Furthermore, the ML predictive model 375 may be trained to apply different weighting factors for different configuration parameters to determine a verdict for each incoming configuration parameter and/or an aggregate of weighted configuration parameters for classifying of the malware detection system 120₁ providing the meta-information 130.

According to one embodiment of the disclosure, each set of training data 364 includes normalized, heuristic data associated with one or more configuration parameters, where the training data 364 is labeled to correspond to a particular classification of a plurality of classifications. Based on the foregoing, each classification of the training data 364 may correspond to a different aggregation of configuration parameter values as different hardware profiles may be associated with different normalized, heuristic data operating as a baseline and/or different weighting factors assigned to configuration parameter values for determining a verdict for each incoming configuration parameter value and an aggregate of weighted configuration parameter values.

The machine learning modeling logic 365 processes the ML training model 370 using the labeled training data 362 as well as unlabeled training data 363 to produce the updated predictive model 375. For instance, using the labeled training data 362, the ML training model 370 continues to update and improve the detection accuracy of the ML training model 370 until a prescribed accuracy (e.g., 90% accuracy) is achieved. Thereafter, the ML training model 370 is released for initial testing as the ML predictive model 375, and based on continued reliable testing of the ML predictive model 375, the ML predictive model 375 is utilized by the system configuration optimization engine 150 for determining whether certain malware detection systems are operating at an optimal performance level or a non-optimal performance level. Thereafter, the ML training model 370 (corresponding to the current ML predictive model) continues further training to improve operability of the ML predictive model 375.

The system health reporting logic 380 is adapted to receive the incoming meta-information and utilize the ML predicted model 375 in (1) determining whether the malware detection system is operating at an optimal performance level or not, and (2) determining what configuration parameters are modifiable in order for the malware detection system 120₁ to be adjusted to operate at the optimal level. The system health reporting logic 380 may perform a number of operations iteratively by modifying different configuration parameters and analyzing the results of the modification to determine whether certain configuration parameter values provided by the meta-information gravitate toward the optimal performance level.

For instance, the system health reporting logic 380 may detect that an “over-utilized” processor utilization level (e.g., exceeding a first threshold such as a percentage of processing capacity exceeding 80% utilization where optimal utilization resides within 60%-80% range), and thus, mimic activation of additional processor cores in order to determine whether the activation of a single processor core would be sufficient to reduce the processor utilization level back to an optimal performance level (e.g., operating utilization normalized to reside within 60%-80% range). Hence, the system health reporting logic 380 performs behavioral analysis in accordance with the hardware profile in order to determine that the alteration of certain configuration parameter values is sufficient to return the malware detection system back to an optimal operating range. Besides percentage of processing capacity, the utilization level may be directed to the time spent on idle tasks (e.g., optimal performance level corresponds to a prescribed percentage range of processing time being spent on idle tasks where over-utilization exceeds the prescribed percentage range) or the type and/or amount of computing tasks being performed for a determined measure of time such as per second, hour, day or the like (e.g. the optimal performance level may correspond to a prescribed range of computing tasks performed according to the determined measure of time, where over-utilization exceeds the prescribed computing task range).

Similarly, the system health reporting logic 380 may detect that an “under-utilized” processor utilization level (e.g., utilizing falling below a second threshold such as 30% utilization where utilization normalized to reside within 60%-80% range), and thus, mimic deactivation of a processor core if multiple processor cores are active in order to determine whether the deactivation of a single processor core would be sufficient to increase the processor utilization level back to the optimal performance level. Furthermore, processor under-utilization may be detected where the processing time being spent on idle tasks falls below the prescribed processing range or the number of computing tasks performed over the measured unit of time falls below the prescribed computing task range.

Referring to FIG. 4, a first exemplary embodiment of an operation flow between a cybersecurity system 110₁ and the system configuration optimization engine 150 is shown. Herein, both the malware detection systems 120₁ and 120₂ communicate with a cloud service 400, which operates as a datacenter that aggregates meta-information 130 received from malware detection systems 120₁ via a first message 410 and meta-information 1302 received from malware detection systems 120₂ via a second message 415. As shown, one embodiment of the first message 410 may include an identifier of the malware detection system 120₁ (e.g., System ID 420), an optional timestamp 425, and statistics 135 associated with one or more configuration parameters representing the system metrics, such as hardware utilization statistics 430 (e.g., processor or memory-based statistics), virtual machine utilization statistics 435 (e.g., VM-based statistics) and/or software utilization statistics 440 (e.g., process-based statistics) as described above.

During aggregation, according to one embodiment of the disclosure, the unstructured data within the meta-information 130 may be formatted and placed into a prescribed data structure. Otherwise, the meta-information 130, including structured and/or unstructured data, may be provided to the system configuration optimization engine 150. According to one embodiment of the disclosure, the cloud service 400 controls delivery of the meta-information 130 (e.g., “push” delivery) while, according to another embodiment, the system configuration optimization engine 150 controls delivery of the meta-information 130 (e.g., “pull” delivery).

As shown, the system configuration optimization engine 150 includes the parser 350, which features a plurality of sub-parsers 355 including the first sub-parser 356 and the second sub-parser 357. As described above, the first sub-parser 356 is configured to parse structured data contained in the meta-information 130 in order to recover one or more configuration parameter values. The recovered configuration parameter values are used by the ML predictive model 375, being processed by the ML modeling logic (not shown), in determining a performance level at which the malware detection system 120₁ is currently operating. Additionally, the ML predictive model 375, being processed by the ML modeling logic 365 of FIG. 3, determines the performance level (e.g. values) for each of the recovered configuration parameter values.

Additionally, the second sub-parser 357 is configured to parse unstructured data for relevant information (e.g., analysis of text strings such as lines of a text file). The “relevant” information includes information associated with a monitored event, where the information may be obtained from keyword searches, as described above. The relevant information may be used by the ML modeling logic in determining, independent or in combination with the recovered configuration parameter values, the performance level at which the malware detection system 120₁ is currently operating.

More specifically, the ML modeling logic 365 is applied to the configuration parameter value(s) and/or relevant information are provided to the ML modeling logic 365. The ML predictive model 375 is generated as a result of the ML modeling logic performing “training” operations on the ML training model using the training data as described in FIG. 3. Herein, the ML predictive model 375 determines the hardware profile of a source of the configuration parameter value(s) and/or relevant information, where the hardware profile may influence what normalized, heuristic data is referenced in the classification of the source. For instance, using the ML predictive model 375, the ML modeling logic analyzes portions of the meta-information 130 (e.g., recovered configuration parameter statistics and/or the relevant information) to classify the malware detection system 120₁ (e.g., over-utilized performance level, optimal performance level, or under-utilized performance level).

As described above, the classification operations are dependent on detected hardware profile for the malware detection system 120₁ and the content of the configuration parameters supplied by the meta-information 130. For example, the ML predictive model 375 may apply prescribed weightings to the configuration parameter values, where the aggregate of the weighted values is used to determine whether the malware detection system is operating at an optimal performance level, or is operating at a non-optimal performance level (e.g., over-utilized where processor utilization exceeds a first prescribed percentage and/or memory available falls below a first prescribed byte size or under-utilized where processor utilization falls below a second prescribed percentage and/or memory available exceeds a second prescribed byte size).

The system health reporting logic 380 is adapted to receive the incoming meta-information and utilize the ML predicted model 375 to determine (1) whether the malware detection system is operating at an optimal performance level or not, and (2) determine what configuration parameters are modifiable for adjusting operability of the malware detection system 120₁ to operate at the optimal performance level. The system health reporting logic 380 may iteratively modify certain configuration parameters based on what configuration parameters are negatively influencing performance, and using the resulting affects as feedback to adjust the next iteration so as to tune in steps to a value or values that produces an optimal performance level the configuration of the malware detection system 120₁. This analysis may be performed through behavioral analysis of a virtual machine configured to accordance with a determined hardware profile or through heuristics based on prior configuration parameter adjustments. For instance, the system health reporting logic 380 may detect that an “over-utilized” processor utilization level, and thus, mimic activation of additional processor cores in order to determine whether the activation of a single processor core would be sufficient to reduce the processor utilization level back to an optimal performance level (e.g., operating utilization normalized to reside within 60%-80% range). Hence, the system health reporting logic 380 performs behavioral analysis in accordance with the hardware profile in order to determine that the alteration of certain configuration parameters is sufficient to return the malware detection system back to an optimal performance level.

As described above, the system configuration optimization engine 150 determines, using the ML predictive model 375, whether the meta-information 130 identifies the malware detection system 120₁ as running in an over-utilized performance level, an optimal performance level, or an under-utilized performance level. In response to determining that the malware detection system 120₁ operates at an over-utilized performance level, the system health reporting logic 380 determines which configuration parameters may be altered in order to return the malware detection system 120 back to its optimal performance level. This may involve an increase (or reduction) in active processor cores, an increase (or reduction) in active virtual machine instances, an increase (or reduction) in memory usage, or the like.

Upon completion of the analysis of the meta-information 130 supported by the malware detection system 120₁, the system health reporting logic 380 generates a system health message 450, which is provided to the malware detection system 120₁ that supplied the meta-information 130. Herein, the system health message 450 may include (i) an identifier of the malware detection system supplying the analyzed meta-information; (ii) the performance level for the malware detection system; (iii) the performance level determinations for some or all of the plurality of configuration parameters; and/or (iv) one or more modified configuration parameter values that are used by the malware detection system to adjust its configuration to remain in or return to its optimal performance level. The system health message 450 is consistent with the configuration modification message 170 of FIG. 1, including one or more modified configuration parameter values 175.

As shown, the system health reporting logic 380 may provide the system health message 450 to a customer support service 460. The customer support service 460 may automatically analyze the contents of the system health message 450 and generate subsequent communications 465 (e.g., via electronic mail, text, automated audio, signaling to monitoring and reporting logic 470, etc.) to advise the customer as to proposed modifications to the cybersecurity system. Furthermore, the customer support service 460 may provide a portion of the system health message 450, such as modified configuration parameter values, to a targeted malware detection system. The modified configuration parameter values may be selected to perform system modifications directed to (i) hardware characteristic changes (e.g., number of active processor cores, network connector types or functionality such as activation of wireless transceivers supporting different wireless frequencies, memory capacity thresholds, etc.), (ii) operational setting changes (e.g., OS setting changes, number of active VMs, VM utilization, additional systems or services available for purchase to improve operability of the cybersecurity system, etc.), and/or (iii) software characteristic changes (e.g., number of active processes, applications utilized by the active processes, etc.) or the like. Additionally, or in the alternative, the system health reporting logic 380 may provide the system health message 450 to the monitoring and reporting service 470. The monitoring and reporting service 470 generates automatically, without user interaction, a report (e.g., information for generation of one or more display screens, a printed report, etc.). The report may be provided to a management console or an administrative interface of a targeted malware detection system. Similarly, the portion of the system health message 450, including modified configuration parameter values, may be provided to the targeted malware detection system (e.g., malware detection system 120₁) via the administrative interface 230 of FIG. 2 or via a management console to which each cybersecurity system and/or malware detection system as access.

As shown in FIG. 5, based on the contents of the system health message 450, the monitoring and reporting service 470 generates a report that highlights information associated with performance levels for each of the malware detection systems for each particular customer. Particular types of performance levels (e.g., over-utilized performance levels and/or under-utilized performance levels) may be highlighted to visibly denote a deviation from the optimal performance level. Examples as to how the performance level may be highlighted includes (1) altering location or ordering of at least certain portions of the performance level information to prominently display such information within the report; (2) modifying the font (e.g., color, size, type style, and/or effects) used in conveying some of the malware detection systems operating at non-optimal performance levels; (3) placement of one or more images proximate to a listing of certain types of performance levels (e.g., optimal, non-optimal, etc.); and (4) placement in a special window or windows associated with listings of certain types of performance levels.

Referring still to FIG. 5, an exemplary embodiment of an interface display screen 500 produced by the monitoring and reporting service 470 of FIG. 4 that provides an interactive dashboard is shown. Herein, rendered by the monitoring and reporting service 470, the display screen 500 features a first display area 510 that illustrates information directed to the performance level determined for malware detection systems deployed on-premises at a customer site. Multiple highlighting techniques are shown in display screen 500, although it is contemplated that any one or more highlighting technique may be conducted for a particular display.

More specifically, according to one embodiment of the disclosure, the display area 510 displays a plurality of entries 520₁-520_R (R>1, R=3 for this embodiment) that provide information directed to performance levels of the malware detection systems 120₁-120_M for each customer. As shown, each row of entries (e.g., 520₁) rendered by the display logic comprises a plurality of fields, including one or more of the following: (1) a first field 530 including an identifier of the malware detection system; (2) a second field 532 including a timestamp that identifies when an analysis of the performance level for the malware detection system was conducted by the system configuration optimization engine 150; and/or (3) a third field 534 including the predicted performance level determined for the malware detection system by the system configuration optimization engine 150. The display area 510 may include additional fields to provide more details directed to the malware detection systems associated with a particular customer, including a fourth field 536 that lists a host address for the corresponding malware detection system, and/or a fifth field 538 that lists a hardware profile for the corresponding malware detection system.

Herein, the fields 530, 532, and 534 associated with malware detection systems operating at non-optimal performance levels may warrant heightened scrutiny level, namely information is displayed more prominently than those fields associated with malware detection systems operating at optimal performance levels for example. This allows a network administrator to more quickly and easily determine malware detection systems that may need re-configuration to improve system operability.

As an example, as a highlighting technique illustrated for the first field 530, the font associated with the malware detection systems operating at non-optimal performance levels (SYSTEM 1; SYSTEM 3) may be displayed differently than the font associated with the host names for malware detection system operating at an optimal performance level (SYSTEM 2). Alternatively, or in addition to the font changes in display, the highlighting technique may be accomplished by ordering malware detection systems operating at non-optimal performance levels (SYSTEM 1; SYSTEM 3) at the top of a listing while any malware detection systems operating at optimal performance levels (SYSTEM 2) are ordered toward the bottom of the listing. As another alternative embodiment, although not shown, a single display screen may produce two areas, where a first area includes the malware detection systems operating at non-optimal performance levels (SYSTEM 1; SYSTEM 3) while a second area includes one or more malware detection systems operating at optimal performance levels (SYSTEM 2).

As further granularity of the operability of the malware detection system under analysis may be needed, according to one embodiment, selection of a field associated with a targeted malware detection system (e.g., performance level field 534 determined for the malware detection system by the system configuration optimization engine 150) allows the user to visualize the performance level of each individual configuration parameter, as shown in FIG. 6.

Referring to FIG. 6, it is contemplated that selection of a particular entry (e.g., third entry 534 including the performance level represented by an underlined portion) may enable the network administrator to obtain more detailed information of the configuration parameter values that resulted in determining whether a certain malware detection system is operating at an optimal performance level or a non-optimal performance level. For instance, as shown in FIG. 6, by selecting a particular entry (e.g., third entry 534), the administrator may be able to uncover the predictive results that lead to the predicted performance level.

According to one embodiment of the disclosure, the predictive results may include each statistic associated with a configuration parameter 610 supplied by the malware detection system 120₁ as part of the meta-information 130, the value 620 associated with each current configuration parameter received from the malware detection system 120₁, and the performance level 630 determined for that particular configuration parameter (i.e., “over-utilized,” “optimal,” and “under-utilized”). Also, as an optional feature, the weighting 640 allocated for each configuration parameter may be displayed with the performance level determination along with the normalized optimal range 650 based on heuristic data for the particular hardware profile.

Referring to FIG. 7, an exemplary embodiment of the logical operations performed by the malware detection systems 120₁ (deployed within the cybersecurity system 110₁) and the system configuration optimization engine 150 is shown. Herein, on a periodic basis and/or an aperiodic basis, the configuration analysis logic 265 collects configuration data associated with the malware detection system 120₁. According to this embodiment, the OS statistics module 266 is configured to collect hardware utilization statistics 700, which may include current operating state information associated with the processor(s) (processor statistics 705) and memory (memory statistics 710). Examples of the processor statistics 705 may include, but are not limited to processor utilization and/or the number of processor cores active. Examples of the memory statistics 710 may include, but are not limited to available memory (size) and/or average (disk) access speed.

As shown in FIG. 7, the configuration analysis logic 265 further includes the VM statistics module 267 and the application statistics module 268. The VM statistics module 267 may be configured to collect VM utilization statistics such as the current virtual operating state associated with a virtual processor (VM processor statistics 715) and/or a virtual memory (virtual memory statistics 720). Examples of the VM processor statistics 715 may include, but are not limited to number of maximum VM instances available or running, or number of concurrent VMs running, or queue length of objects waiting for dynamic analysis, or guest CPU utilization of each VM. Examples of the virtual memory statistics 720 may include, but are not limited to virtual memory available (size). The application statistics module 268 may be configured to collect software statistics such as what processes are running 725 and event log outlining operations of the processes (not shown).

Thereafter, the configuration analysis logic 265 may temporarily store these statistics as the meta-information associated with the malware detection system 120₁ and the meta-information 130 is made available to the system configuration optimization engine 150. For this embodiment, the meta-information 135 is transmitted to the system configuration optimization engine 150 and evaluated, using a predictive model developed based on heuristic data gathered from experiential knowledge of operational meta-information of known misconfigured and optimally configured systems. For instance, when processed by the ML modeling logic, the configuration parameters (and/or groups of configuration parameters) are compared to heuristic data associated with optimal and non-optimal performance levels associated with the respective configuration parameters (and/or groups of configuration parameters) to determine whether the configuration parameter value (and/or groups of configuration parameter values) falls within a prescribed range as determined by the heuristic data.

For those configuration parameter values falling outside of the optimal performance level, the system configuration optimization engine 150 determines (what configuration parameter values are modifiable for adjusting operability of the malware detection system 120₁ to operate at the optimal performance level. The system configuration optimization engine 150 iteratively modifies certain configuration parameter values based on what configuration parameter values are negatively influencing performance by the malware detection system 120₁ and analyzes the potential effects of a similar adjustment to the configuration of the malware detection system 120₁.

As shown in FIG. 7, the system configuration optimization engine 150 determines which configuration parameter values may be altered in order to return the malware detection system 120₁ back to its optimal performance level. More specifically, the system configuration optimization engine 150 may perform hardware optimization modifications 730 such as adjusting (i.e., increasing or reducing) the number of active processor cores or adjusting the number of log events maintained by the malware detection system 120₁ (to increase or reduce processor utilization). Additionally, or in the alternative, the system configuration optimization engine 150 may perform VM optimization modifications 735 such as adjusting (i.e., increasing or reducing) the number of active virtual machine instances. Likewise, additionally or in the alternative, the system configuration optimization engine 150 may perform software optimization modifications 740 such as enabling or disabling certain software features or adjusting (i.e., increasing or reducing) resources available to a particular software process.

Referring now to FIG. 8, an exemplary embodiment of a flowchart illustrating operations of the system configuration optimization engine of FIG. 1, FIG. 3 and FIG. 4 is shown. Initially, meta-information including configuration parameters directed to the operability of a malware detection system is received (block 800). From the meta-information, the hardware profile of the malware detection system is determined to identify a desired operating configuration when analyzing the configuration parameter values associated with the malware detection system (block 810). The incoming meta-information is analyzed against the heuristic configuration data in accordance with the hardware profile of the malware detection system in order to determine whether or not the incoming meta-information indicates that the malware detection system is operating at an optimal performance level (blocks 820, 830 and 840).

After the determination of the performance level of the malware detection system, the system configuration optimization engine determines which configuration parameter values associated with the incoming meta-information are modifiable to adjust the configuration (and operating state) of the malware detection system (block 850). Thereafter, the system configuration optimization engine may be configured to perform, within a virtualized environment representative of the hardware profile of the malware detection system, iterative adjustments of different configuration parameter values to determine whether such adjustments allow the current performance level to remain in or return to its optimal performance level (block 860). The performance levels of the malware detection system and configuration parameter values may be provided for display (block 870). The modified configuration parameter values may be automatically returned to the malware detection system for reconfiguring the malware detection system (blocks 880 and 890).

Referring to FIG. 9, a second exemplary embodiment of an operation flow between the cybersecurity system 110₁ and the system configuration optimization engine 150 is shown. Herein, as shown, the malware detection systems 120₁ is in communication with the system configuration optimization engine 150 via a cloud service 900. The cloud service 900 is configured to aggregate incoming meta-information 130 from different malware detection systems 120₁-120_N of the cybersecurity system 110₁ and control the return of modified configuration parameter values and the updating of the predictive model 375.

More specifically, the malware detection systems 120₁, in response to a detected event or timeout, obtains meta-information associated with the current operating state of the malware detection system 120₁. Herein, the meta-information 130 includes hardware utilization statistics, VM utilization statistics, and/or software utilization statistics as described above.

As shown, the system configuration optimization engine 150 includes the parser 350, to recover one or more the configuration parameter values contained in the meta-information 130. The statistics are used by the ML predictive model 375, being processed by the ML modeling logic 365, in determining a performance level at which the malware detection system 120₁ is currently operating. Additionally, the ML predictive model 375, being processed by the ML modeling logic 365, determines the performance level for each of the configuration parameter values pertaining to the statistics. The ML predictive model 375 is repeatedly updated and configured as a result of the ML modeling logic 365 performing “training” operations on the training data 364.

Herein, the system configuration optimization engine 150 determines the hardware profile of a source of the configuration parameter value(s), where the hardware profile may influence what normalized, heuristic data is referenced in the classification of the source. For instance, using the ML predictive model 375, the ML modeling logic 365 analyzes portions of the meta-information 130 (e.g., statistics associated with configuration parameters) to classify the malware detection system 120₁ (e.g., over-utilized performance level, optimal performance level, or under-utilized performance level). Furthermore, the system configuration optimization engine 150 determines what configuration parameter values are modifiable for adjusting operability of the malware detection system 120₁ to operate at the optimal performance level. Thereafter, the system configuration optimization engine 150 iteratively modifies a subset of these configuration parameter values to determine what configuration modifications of the malware detection system 120₁ will maintain the system at an optimal performance level.

More specifically, the system configuration optimization engine 150 determines, using the ML predictive model 375, whether the meta-information 130 identifies the malware detection system 120₁ as running in a non-optimal performance level and the configuration modification necessary to return the malware detection system 120₁. In response to determining that the malware detection system 120₁ operates at an over-utilized performance level, the system configuration optimization engine 150 determines one or more configuration parameter values that, if altered, improve the performance level of the malware detection system 120₁ and returns update information 900 that would cause the configuration modification at the malware detection system 120₁ to occur in real-time. Updates 910 to the ML predictive model 375 are provided to the malware detection system 120₁ to analyze the performance level of the malware detection system 120₁ locally and in real-time.

In the foregoing description, the invention is described with reference to specific exemplary embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention as set forth in the appended claims.

Claims

1. A computerized method for reconfiguring one or more malware detection systems each performing cybersecurity analyses on incoming data, the method comprising: receiving meta-information including metrics associated with the one or more malware detection systems;determining whether each of the one or more malware detection systems is operating at an optimal performance level by at least determining a correlation between the metrics and a baseline configuration being metrics gathered through experiential knowledge of meta-information associated with known malware detection systems, wherein the baseline configuration comprises utilization statistics including one or more of (i) hardware utilization statistics, (ii) virtual machine utilization statistics, or (iii) software utilization statistics; andgenerating results, based on the correlation between the metrics and the baseline configuration, provided as feedback to the one or more malware detection systems to update one or more configuration parameters of the one or more malware detection systems.

2. The computerized method of claim 1, wherein the metrics include statistics associated with the one or more configuration parameters.

3. The computerized method of claim 1, wherein the metrics include events monitored during operations of a malware detection system of the one or more malware detection systems.

4. The computerized method of claim 1, wherein each of the one or more malware detection systems being deployed as a cloud-based service and the optimal performance level associated with a malware detection system of the one or more malware detection systems corresponds to a preferred operating state for the malware detection system as measured by any one or more of the hardware utilization statistics, the virtual machine utilization statistics or the software utilization statistics associated with the malware detection system.

5. The computerized method of claim 1 further comprising: sending an alert message that requires an action by an administrator before one or more modified configuration parameters are provided to a malware detection system of the one or more malware detection systems to update the one or more configuration parameters of the malware detection system.

6. The computerized method of claim 1, wherein each of the one or more configuration parameters includes information that partially controls an operating state of a resource deployed within a malware detection system of the one or more malware detection systems.

7. The computerized method of claim 6, wherein a configuration parameter of the one or more configuration parameters is directed to virtual machine characteristics including a virtual machine utilization level.

8. The computerized method of claim 6, wherein the determining of the correlation between the metrics and the baseline configuration is conducted by a predictive model in operation within a system configuration optimization engine.

9. The computerized method of claim 8, wherein the correlation conducted by the predictive model comprises an evaluation of configuration parameter values being part of the metrics with configuration parameter values pertaining to past configurations that are associated with either (i) non-optimal performance levels, (ii) optimal performance levels, or (iii) a combination of non-optimal performance levels and optimal performance levels.

10. The computerized method of claim 8, wherein the determining whether each of the one or more malware detection systems is operating at the optimal performance level comprises determining, by the predictive model, whether a malware detection system of the one or more malware detection system is operating at the optimal performance level that represents a preferred operating state for the malware detection system performing cybersecurity analyses.

11. The computerized method of claim 8, wherein the predictive model is produced by at least training a machine learning model that includes applying different weighting factors to at least one of the one or more configuration parameters.

12. The computerized method of claim 1, wherein the utilization statistics are represented by at least one of (i) a threshold, (ii) a range, (iii) a number, or (iv) an operational bound.

13. The computerized method of claim 1, wherein the one or more configuration parameters include information to at least partially control an operating state of software operating within a malware detection system of the one or more malware detection systems.

14. The computerized method of claim 13, wherein the one or more configuration parameters includes a first configuration parameter directed to an operational setting of the malware detection system.

15. The computerized method of claim 14, wherein the operational setting includes one or more of (i) virtual machine characteristics including a number of active virtual machines used by the malware detection system or (ii) software characteristics including a number of active processes or a number of applications utilized by the active processes.

16. The computerized method of claim 13, wherein the baseline configuration associated with the malware detection system is selected based, at least in part, on a hardware profile for the malware detection system, along with other factors (e.g., threat landscape confronting the customer (for instance, as indicated by the industry protected by the malware detection system), subscription or customer type, etc.).

17. The computerized method of claim 13, wherein the baseline configuration associated with the malware detection system is selected based, at least in part, on a threat landscape confronting a customer associated with the one or more malware detection systems, the threat landscape reflects one or more factors including an industry protected by the malware detection system, a subscription type utilized by the customer, or a type of customer.

18. The computerized method of claim 1, wherein the one or more malware detection systems are deployed within a public cloud network.

19. The computerized method of claim 18, wherein a system configuration optimization engine is configured to conduct the receiving, determining and generating operations is deployed within the public cloud network.

20. The computerized method of claim 1, wherein the optimal performance level corresponds to or is associated with a preferred operating state of a malware detection system of the one or more malware detection system as measured by the hardware utilization statistics, the virtual machine utilization statistics, and/or the software utilization statistics.

21. The computerized method of claim 1, wherein the determining whether each of the one or more malware detection systems is operating at the optimal performance level comprises a determination as to a degree of correlation between the metrics associated with the received meta-information and metrics associated with the baseline configuration including metrics associated with one or more of (i) known misconfigured malware detection systems, (ii) known optimal malware detection systems, or (iii) any combination of known misconfigured malware detection systems and known optimal malware detection systems.

22. The computerized method of claim 1, wherein the determining whether each of the one or more malware detection systems is operating at the optimal performance level comprises determining a performance level for at least a malware detection system of the one or more malware detection systems by conducting operations on metrics provided by the malware detection system as part of the meta-information, the operations include assigning weighting to different metrics of the metrics provided by the malware detection system, the weighting to adjust the performance level toward one or more configuration parameters associated with the metrics that are more applicable to representing the performance level of the malware detection system when correlating with the baseline configuration.

23. The computerized method of claim 1, wherein the generating of the results to be provided as feedback to the one or more malware detection systems is conducted to reduce a number of false negatives.

24. The computerized method of claim 1, wherein the hardware utilization statistics includes at least one of processor utilization or an amount or degree of available memory for storage.

25. The computerized method of claim 1, wherein the virtual machine utilization statistics includes information directed to a degree of utilization of one or more virtual machines operating within a selected malware detection system or information directed to identifying a number of virtual machines activated and operational within a selected malware detection system.

26. The computerized method of claim 1, wherein the software utilization statistics include (i) information to determine what processes are running within a selected malware detection system or (ii) statistics associated with the processes or queue length.

27. The computerized method of claim 1, wherein the one or more malware detection systems are part of a cloud service.

28. The computerized method of claim 1 is performed in a cloud service.

29. The computerized method of claim 1, wherein the correlation between the metrics and the baseline configuration is conducted to determine whether a malware detection system of the one or more malware detection systems is operating at a non-optimal performance level, the non-optimal performance level indicating that the malware detection system is operating outside of a desired operating configuration.

30. The computerized method of claim 1, wherein the feedback includes one or more modified configuration parameter values that differ from values of the one or more configuration parameters received as part of the meta-information.

31. The computerized method of claim 30 further comprising: updating the one or more configuration parameters of the one or more malware detection systems with the one or more modified configuration parameter values.

32. The computerized method of claim 31, wherein the updating of the one or more configuration parameters with the one or more modified configuration parameter values occurs in response to receipt of an approval from an administrator.

33. The computerized method of claim 1, wherein the one or more malware detection system includes a first malware detection system and the baseline configuration is based on a profile of the first malware detection system.

34. A system for detecting a cyber-attack comprising: one or more processors;a memory communicatively coupled to the one or more processors, the memory comprises(i) a parser that, upon execution-by the one or more hardware processors, receives incoming data from a network device and recovers meta-information associated with one or more configuration parameters,(ii) a machine learning-modeling logic that, upon execution by the one or more processors, conducts training on a machine learning model to produce a predictive model being applied to training data including data associated with the recovered meta-information to determine whether the network device is operating at a non-optimal performance level, and(iii) a system health reporting logic that, upon execution by the one or more processors and based on the recovered meta-information, (a) determines whether the network device is operating at the non-optimal performance level, (b) determines one or more configuration parameter values associated with the one or more configuration parameters that, if modifiable, readjusts operability of the network device from the non-optimal performance level to an optimal performance level, and (c) generates a message including information to alter the one or more configuration parameter values at the network device.

35. The system of claim 34, wherein the parser of the memory comprises a first sub-parser configured to parse structured data of the incoming data to recover the meta-information associated with the one or more configuration parameter values.

36. The system of claim 35, wherein the parser of the memory further comprises a second sub-parser configured to parse unstructured data of the incoming data to recover information subsequent to specific keywords being monitored.

37. The system of claim 35, wherein the recovered meta-information including statistics associated with the one or more configuration parameter values that represent a performance level of the network device.

38. The system of claim 37, wherein the parser that, upon execution by the one or more processors, receives incoming data from the network device being a malware detection system including circuitry to detect and analyze incoming objects propagating into or through a local network for malware.

39. The system of claim 35, wherein the system health reporting logic that, upon execution by the one or more processors, determines whether the network device is operating at the non-optimal performance level based on at least a prescribed degree of correlation between the statistics associated with a first configuration parameter of the one or more configuration parameters and the optimal performance level being a preferred operating state for the network device as measured by hardware utilization statistics.

40. The system of claim 35, wherein the system health reporting logic that, upon execution by the one or more processors, determines whether the network device is operating at the non-optimal performance level based on at least a prescribed degree of correlation between the statistics associated with a first configuration parameter of the one or more configuration parameters and the optimal performance level being a preferred operating state for the network device as measured by virtual machine utilization statistics.

41. The system of claim 35, wherein the system health reporting logic that, upon execution by the one or more processors, determines whether the network device is operating at the non-optimal performance level based on at least a prescribed degree of correlation between the statistics associated with a first configuration parameter of the one or more configuration parameters and the optimal performance level being a preferred operating state for the network device as measured by software utilization statistics.

42. The system of claim 34, wherein the system health reporting logic, upon execution by the one or more processors, performs iterative operations including modifying at least one configuration parameter value and analyzing predicted changes in performance level of the source in returning the network device to the optimal performance level.

43. The system of claim 8 further comprising an administrative interface operating as is a portal that allows an administrator, after credential exchange and authentication, to access and update logic stored within the memory, the portal allows for accessing and updating logic by at least modifying (i) one or more triggering events or timeout parameters associated with an event/timeout monitoring logic executed by the one or more processors, or (ii) code configured to detect a presence of malware within an object or its association with a cyberattack, or (iii) operational settings associated with the malware detection system, or (iv) software associated with the malware detection system.

44. The system of claim 34, wherein the machine learning modeling logic, upon execution by the one or more processors, conducts training on the machine learning model to produce the predictive model using labeled training data and unlabeled training data.

45. The system of claim 44, wherein the machine learning modeling logic, upon execution by the one or more processors, conducts the supervised training on the machine learning model using the labeled training data to initially train the machine learning model to improve detection accuracy of the machine learning model in determining whether the network device is operating at a non-optimal performance level until a prescribed accuracy is achieved.

46. A system for detecting a cyber-attack comprising: one or more processors;a memory communicatively coupled to the one or more processors, the memory comprises(i) a parser that, upon execution by the one or more processors, receives incoming data from a network device and recovers meta-information associated with one or more configuration parameters and determines a hardware profile of the network device,(ii) a machine learning modeling logic that, upon execution by the one or more processors, processes a machine learning predictive model that determines whether the network device is operating at a non-optimal performance level or an optimal performance level based on the recovered meta-information, and(iii) a system health reporting logic that, upon execution by the one or more processors and utilizes the machine learning predictive model, (a) determines whether the network device is operating at the non-optimal performance level, (b) determines one or more configuration parameter values associated with the one or more configuration parameters that, if modifiable, readjusts operability of the network device from the non-optimal performance level to an optimal performance level, and (c) generates a message including information to alter the one or more configuration parameter values at the network device.

47. The system of claim 46, wherein the system health reporting logic that, upon execution by the one or more processors, further generates an update of the machine learning predictive model to enable the network device to analyze performance level locally and in real time.

48. The system of claim 46, wherein the machine learning modeling logic being configured to conduct training on a machine learning model using training data to produce the machine learning predictive model, the training data includes information associated the meta-information.

49. The system of claim 48, wherein the information associated with the meta-information includes normalized, heuristic data pertaining to the configuration parameter values.

50. The system of claim 48, wherein the machine learning modeling logic is configured to evaluate operability of the machine learning model by applying different weighting factors for different configuration parameter values of the configuration parameter values received from the parser.

51. The system of claim 46, wherein the machine learning modeling logic is configured to evaluate operability of the machine learning model based on the training data including at least configuration parameter values associated a machine learning model operating on the system health reporting logic.