Patent Application
20230116423
The present disclosure relates generally to computer networks, and, more particularly, to a secure microphone agent.
Voice controls are becoming increasingly ubiquitous across a variety of use cases. Indeed, many personal computing devices now include voice control functionality. In addition, stand-alone voice control devices are also increasing in popularity for use in home automation and as personal voice assistants.
With the proliferation of voice controls also comes increasing threats of malicious use. In a simple case, an unauthorized user may purposely issue a voice command to create harm. In more sophisticated attacks, malware can cause an infected device to issue voice commands to a voice control device or cause an infected voice control device to perform an unwanted action.
The embodiments herein may be better understood by referring to the following description in conjunction with the accompanying drawings in which like reference numerals indicate identically or functionally similar elements, of which:
According to one or more embodiments of the disclosure, a device extracts a voice command from audio data captured by a microphone. The device uses a semantic reasoning engine, to determine a goal of the voice command. The device determines that the goal of the voice command is consistent with prior voice commands issued to the device. The device raises an alert when the goal of the voice command is inconsistent with prior voice commands issued to the device.
A computer network is a geographically distributed collection of nodes interconnected by communication links and segments for transporting data between end nodes, such as personal computers, cellular phones, workstations, or other devices, such as sensors, etc. Many types of networks are available, with the types ranging from local area networks (LANs) to wide area networks (WANs). LANs typically connect the nodes over dedicated private communications links located in the same general physical location, such as a building or campus. WANs, on the other hand, typically connect geographically dispersed nodes over long-distance communications links, such as common carrier telephone lines, optical lightpaths, synchronous optical networks (SONET), or synchronous digital hierarchy (SDH) links, or Powerline Communications (PLC) such as IEEE 61334, IEEE P1901.2, and others. The Internet is an example of a WAN that connects disparate networks throughout the world, providing global communication between nodes on various networks. The nodes typically communicate over the network by exchanging discrete frames or packets of data according to predefined protocols, such as the Transmission Control Protocol/Internet Protocol (TCP/IP). In this context, a protocol consists of a set of rules defining how the nodes interact with each other. Computer networks may be further interconnected by an intermediate network node, such as a router, to forward data from one network to another.
Smart object networks, such as sensor networks, in particular, are a specific type of network having spatially distributed autonomous devices such as sensors, actuators, etc., that cooperatively monitor physical or environmental conditions at different locations, such as, e.g., energy/power consumption, resource consumption (e.g., water/gas/etc. for advanced metering infrastructure or “AMI” applications) temperature, pressure, vibration, sound, radiation, motion, pollutants, etc. Other types of smart objects include actuators, e.g., responsible for turning on/off an engine or perform other actions. Sensor networks, a type of smart object network, are typically shared-media networks, such as wireless or PLC networks. That is, in addition to one or more sensors, each sensor device (node) in a sensor network may generally be equipped with a radio transceiver or other communication port such as PLC, a microcontroller, and an energy source, such as a battery. Often, smart object networks are considered field area networks (FANs), neighborhood area networks (NANs), personal area networks (PANs), etc. Generally, size and cost constraints on smart object nodes (e.g., sensors) result in corresponding constraints on resources such as energy, memory, computational speed and bandwidth.
In some implementations, a router or a set of routers may be connected to a private network (e.g., dedicated leased lines, an optical network, etc.) or a virtual private network (VPN), such as an MPLS VPN utilizing a Service Provider network, via one or more links exhibiting very different network and service level agreement characteristics. For the sake of illustration, a given customer site may fall under any of the following categories:
Servers 152-154 may include, in various embodiments, a network management server (NMS), a dynamic host configuration protocol (DHCP) server, a constrained application protocol (CoAP) server, an outage management system (OMS), an application policy infrastructure controller (APIC), an application server, etc. As would be appreciated, network 100 may include any number of local networks, data centers, cloud environments, devices/nodes, servers, etc.
In some embodiments, the techniques herein may be applied to other network topologies and configurations. For example, the techniques herein may be applied to peering points with high-speed links, data centers, etc.
In various embodiments, network 100 may include one or more mesh networks, such as an Internet of Things network. Loosely, the term “Internet of Things” or “IoT” refers to uniquely identifiable objects (things) and their virtual representations in a network-based architecture. In particular, the next frontier in the evolution of the Internet is the ability to connect more than just computers and communications devices, but rather the ability to connect “objects” in general, such as lights, appliances, vehicles, heating, ventilating, and air-conditioning (HVAC), windows and window shades and blinds, doors, locks, etc. The “Internet of Things” thus generally refers to the interconnection of objects (e.g., smart objects), such as sensors and actuators, over a computer network (e.g., via IP), which may be the public Internet or a private network.
Notably, shared-media mesh networks, such as wireless or PLC networks, etc., are often deployed on what are referred to as Low-Power and Lossy Networks (LLNs), which are a class of network in which both the routers and their interconnect are constrained: LLN routers typically operate with constraints, e.g., processing power, memory, and/or energy (battery), and their interconnects are characterized by, illustratively, high loss rates, low data rates, and/or instability. LLNs are comprised of anything from a few dozen to thousands or even millions of LLN routers, and support point-to-point traffic (between devices inside the LLN), point-to-multipoint traffic (from a central control point such at the root node to a subset of devices inside the LLN), and multipoint-to-point traffic (from devices inside the LLN towards a central control point). Often, an IoT network is implemented with an LLN-like architecture. For example, as shown, local network 160 may be an LLN in which CE-2 operates as a root node for devices/nodes 10-16 in the local mesh, in some embodiments.
In contrast to traditional networks, LLNs face a number of communication challenges. First, LLNs communicate over a physical medium that is strongly affected by environmental conditions that change over time. Some examples include temporal changes in interference (e.g., other wireless networks or electrical appliances), physical obstructions (e.g., doors opening/closing, seasonal changes such as the foliage density of trees, etc.), and propagation characteristics of the physical media (e.g., temperature or humidity changes, etc.). The time scales of such temporal changes can range between milliseconds (e.g., transmissions from other transceivers) to months (e.g., seasonal changes of an outdoor environment). In addition, LLN devices typically use low-cost and low-power designs that limit the capabilities of their transceivers. In particular, LLN transceivers typically provide low throughput. Furthermore, LLN transceivers typically support limited link margin, making the effects of interference and environmental changes visible to link and network protocols. The high number of nodes in LLNs in comparison to traditional networks also makes routing, quality of service (QoS), security, network management, and traffic engineering extremely challenging, to mention a few.
The network interfaces 210 include the mechanical, electrical, and signaling circuitry for communicating data over physical links coupled to the network 100. The network interfaces may be configured to transmit and/or receive data using a variety of different communication protocols. Notably, a physical network interface 210 may also be used to implement one or more virtual network interfaces, such as for virtual private network (VPN) access, known to those skilled in the art.
The memory 240 comprises a plurality of storage locations that are addressable by the processor(s) 220 and the network interfaces 210 for storing software programs and data structures associated with the embodiments described herein. The processor 220 may comprise necessary elements or logic adapted to execute the software programs and manipulate the data structures 245. An operating system 242 (e.g., the Internetworking Operating System, or IOS®, of Cisco Systems, Inc., another operating system, etc.), portions of which are typically resident in memory 240 and executed by the processor(s), functionally organizes the node by, inter alia, invoking network operations in support of software processors and/or services executing on the device. These software processors and/or services may comprise a deep fusion reasoning engine (DFRE) process 248, as described herein.
It will be apparent to those skilled in the art that other processor and memory types, including various computer-readable media, may be used to store and execute program instructions pertaining to the techniques described herein. Also, while the description illustrates various processes, it is expressly contemplated that various processes may be embodied as modules configured to operate in accordance with the techniques herein (e.g., according to the functionality of a similar process). Further, while processes may be shown and/or described separately, those skilled in the art will appreciate that processes may be routines or modules within other processes.
DFRE process 248 includes computer executable instructions that, when executed by processor(s) 220, cause device 200 to provide cognitive reasoning services to a network. In various embodiments, DFRE process 248 may utilize machine learning techniques, in whole or in part, to perform its analysis and reasoning functions. In general, machine learning is concerned with the design and the development of techniques that take as input empirical data (such as network statistics and performance indicators) and recognize complex patterns in these data. One very common pattern among machine learning techniques is the use of an underlying model M, whose hyper-parameters are optimized for minimizing the cost function associated to M, given the input data. The learning process then operates by adjusting the hyper-parameters such that the number of misclassified points is minimal. After this optimization phase (or learning phase), the model M can be used very easily to classify new data points. Often, M is a statistical model, and the minimization of the cost function is equivalent to the maximization of the likelihood function, given the input data.
In various embodiments, DFRE process 248 may employ one or more supervised, unsupervised, or self-supervised machine learning models. Generally, supervised learning entails the use of a training large set of data, as noted above, that is used to train the model to apply labels to the input data. For example, in the case of video recognition and analysis, the training data may include sample video data that depicts a certain object and is labeled as such. On the other end of the spectrum are unsupervised techniques that do not require a training set of labels. Notably, while a supervised learning model may look for previously seen patterns that have been labeled as such, an unsupervised model may instead look to whether there are sudden changes in the behavior. Self-supervised is a representation learning approach that eliminates the pre-requisite requiring humans to label data. Self-supervised learning systems extract and use the naturally available relevant context and embedded metadata as supervisory signals. Self-supervised learning models take a middle ground approach: it is different from unsupervised learning as systems do not learn the inherent structure of data, and it is different from supervised learning as systems learn entirely without using explicitly-provided labels.
Example machine learning techniques that DFRE process 248 can employ may include, but are not limited to, nearest neighbor (NN) techniques (e.g., k-NN models, replicator NN models, etc.), statistical techniques (e.g., Bayesian networks, etc.), clustering techniques (e.g., k-means, mean-shift, etc.), neural networks (e.g., reservoir networks, artificial neural networks, etc.), support vector machines (SVMs), logistic or other regression, Markov models or chains, principal component analysis (PCA) (e.g., for linear models), multi-layer perceptron (MLP) artificial neural networks (ANNs) (e.g., for non-linear models), replicating reservoir networks (e.g., for non-linear models, typically for time series), random forest classification, or the like. Accordingly, DFRE process 248 may employ deep learning, in some embodiments. Generally, deep learning is a subset of machine learning that employs ANNs with multiple layers, with a given layer extracting features or transforming the outputs of the prior layer.
The performance of a machine learning model can be evaluated in a number of ways based on the number of true positives, false positives, true negatives, and/or false negatives of the model. For example, the false positives of the model may refer to the number of times the model incorrectly identified an object or condition within a video feed. Conversely, the false negatives of the model may refer to the number of times the model failed to identify an object or condition within a video feed. True negatives and positives may refer to the number of times the model correctly determined that the object or condition was absent in the video or was present in the video, respectively. Related to these measurements are the concepts of recall and precision. Generally, recall refers to the ratio of true positives to the sum of true positives and false negatives, which quantifies the sensitivity of the model. Similarly, precision refers to the ratio of true positives the sum of true and false positives.
According to various embodiments,
In general, a reasoning engine, also known as a ‘semantic reasoner,’ ‘reasoner,’ or ‘rules engine,’ is a specialized form of machine learning software that uses asserted facts or axioms to infer consequences, logically. Typically, a reasoning engine is a form of inference engine that applies inference rules defined via an ontology language. As introduced herein, a DFRE is an enhanced form of reasoning engine that further leverages the power of sub-symbolic machine learning techniques, such as neural networks (e.g., deep learning), allowing the system to operate across the full spectrum of sub-symbolic data all the way to the symbolic level.
At the lowest layer of hierarchy 300 is sub-symbolic layer 302 that processes the sensor data 312 collected from the network. For example, sensor data 312 may include video feed/stream data from any number of cameras located throughout a location. In some embodiments, sensor data 312 may comprise multimodal sensor data from any number of different types of sensors located throughout the location. At the core of sub-symbolic layer 302 may be one or more DNNs 308 or other machine learning-based model that processes the collected sensor data 312. In other words, sub-symbolic layer 302 may perform sensor fusion on sensor data 312 to identify hidden relationships between the data.
At the opposing end of hierarchy 300 may be symbolic layer 306 that may leverage symbolic learning. In general, symbolic learning includes a set of symbolic grammar rules specifying the representation language of the system, a set of symbolic inference rules specifying the reasoning competence of the system, and a semantic theory containing the definitions of “meaning.” This approach differs from other learning approaches that try to establish generalizations from facts as it is about reasoning and extracting knowledge from knowledge. It combines knowledge representations and reasoning to acquire and ground knowledge from observations in a non-axiomatic way. In other words, in sharp contrast to the sub-symbolic learning performed in layer 302, the symbolic learning and generalized intelligence performed at symbolic layer 306 requires a variety of reasoning and learning paradigms that more closely follows how humans learn and are able to explain why a particular conclusion was reached.
Symbolic learning models what are referred to as “concepts,” which comprise a set of properties. Typically, these properties include an “intent” and an “extent,” whereby the intent offers a symbolic way of identifying the extent of the concept. For example, consider the intent that represents motorcycles. The intent for this concept may be defined by properties such as “having two wheels” and “motorized,” which can be used to identify the extent of the concept (e.g., whether a particular vehicle is a motorcycle).
Linking sub-symbolic layer 302 and symbolic layer 306 may be conceptual layer 304 that leverages conceptual spaces. In general, conceptual spaces are a proposed framework for knowledge representation by a cognitive system on the conceptual level that provides a natural way of representing similarities. Conceptual spaces enable the interaction between different type of data representations as an intermediate level between sub-symbolic and symbolic representations.
More formally, a conceptual space is a geometrical structure which is defined by a set of quality dimensions to allow for the measurement of semantic distances between instances of concepts and for the assignment of quality values to their quality dimensions, which correspond to the properties of the concepts. Thus, a point in a conceptual space S may be represented by an n-dimensional conceptual vector v = <d1, ..., di, ..., dn> where di represents the quality value for the ith quality dimension. For example, consider the concept of taste. A conceptual space for taste may include the following dimensions: sweet, sour, bitter, and salty, each of which may be its own dimension in the conceptual space. The taste of a given food can then be represented as a vector of these qualities in a given space (e.g., ice cream may fall farther along the sweet dimension than that of peanut butter, peanut butter may fall farther along the salty dimension than that of ice cream, etc.). By representing concepts within a geometric conceptual space, similarities can be compared in geometric terms, based on the Manhattan distance between domains or the Euclidean distance within a domain in the space. In addition, similar objects can be grouped into meaningful conceptual space regions through the application of clustering techniques, which extract concepts from data (e.g., observations).
Said differently, a conceptual space is a framework for representing information that models human-like reasoning to compose concepts using other existing concepts. Note that these representations are not competing with symbolic or associationistic representations. Rather, the three kinds can be seen as three levels of representations of cognition with different scales of resolution and complementary. Namely, a conceptual space is built up from geometrical representations based on a number of quality dimensions that complements the symbolic and deep learning models of symbolic layer 306 and sub-symbolic layer 302, representing an operational bridge between them. Each quality dimension may also include any number of attributes, which present other features of objects in a metric subspace based on their measured quality values. Here, similarity between concepts is just a matter of metric distance between them in the conceptual space in which they are embedded.
In other words, a conceptual space is a geometrical representation which allows the discovery of regions that are physically or functionally linked to each other and to abstract symbols used in symbolic layer 306, allowing for the discovery of correlations shared by the conceptual domains during concepts formation. For example, an alert prioritization module may use connectivity to directly acquire and evaluate alerts as evidence. Possible enhancements may include using volume of alerts and novelty of adjacent (spatially / temporally) alerts, to tune level of alertness.
In general, the conceptual space at conceptual layer 304 allows for the discovery of regions that are naturally linked to abstract symbols used in symbolic layer 306. The overall model is bi-directional as it is planned for predictions and action prescriptions depending on the data causing the activation in sub-symbolic layer 302.
Layer hierarchy 300 shown is particularly appealing when matched with the attention mechanism provided by a cognitive system that operates under the assumption of limited resources and time-constraints. For practical applications, the reasoning logic in symbolic layer 306 may be non-axiomatic and constructed around the assumption of insufficient knowledge and resources (AIKR). It may be implemented, for example, with a Non-Axiomatic Reasoning System (open-NARS) 310. However, other reasoning engines can also be used, such as Auto-catalytic Endogenous Reflective Architecture (AERA), OpenCog, and the like, in symbolic layer 306, in further embodiments. Even Prolog may be suitable, in some cases, to implement a reasoning engine in symbolic layer 306. In turn, an output 314 coming from symbolic layer 306 may be provided to a user interface (UI) for review. For example, output 314 may comprise a video feed/stream augmented with inferences or conclusions made by the DFRE, such as the locations of unstocked or under-stocked shelves, etc.
By way of example of symbolic reasoning, consider the ancient Greek syllogism: (1.) All men are mortal, (2.) Socrates is a man, and (3.) therefore, Socrates is mortal. Depending on the formal language used for the symbolic reasoner, these statements can be represented as symbols of a term logic. For example, the first statement can be represented as “man ➔[mortal]” and the second statement can be represented as “{ Socrates} ➔man.” Thus, the relationship between terms can be used by the reasoner to make inferences and arrive at a conclusion (e.g., “Socrates is mortal”). Non-axiomatic reasoning systems (NARS) generally differ from more traditional axiomatic reasoners in that the former applies a truth value to each statement, based on the amount of evidence available and observations retrieved, while the latter relies on axioms that are treated as a baseline of truth from which inferences and conclusions can be made.
In other words, a DFRE generally refers to a cognitive engine capable of taking sub-symbolic data as input (e.g., raw or processed sensor data regarding a monitored system), recognizing symbolic concepts from that data, and applying symbolic reasoning to the concepts, to draw conclusions about the monitored system.
According to various embodiments,
In various embodiments, DFRE middleware 402 may also provide services to support semantic reasoning, such as by an AIKR reasoner. For example, as shown, DFRE middleware 402 may include a NARS agent that performs semantic reasoning for structural learning. In other embodiments, OpenCog or another suitable AIKR semantic reasoner could be used.
One or more DFRE agents 404 may interface with DFRE middleware 402 to orchestrate the various services available from DFRE middleware 402. In addition, DFRE agent 404 may feed and interact with the AIKR reasoner so as to populate and leverage a DFRE knowledge graph with knowledge.
More specifically, in various embodiments, DFRE middleware 402 may obtain sub-symbolic data 408. In turn, DFRE middleware 402 may leverage various ontologies, programs, rules, and/or structured text 410 to translate sub-symbolic data 408 into symbolic data 412 for consumption by DFRE agent 404. This allows DFRE agent 404 to apply symbolic reasoning to symbolic data 412, to populate and update a DFRE knowledge base (KB) 416 with knowledge 414 regarding the problem space (e.g., the network under observation, etc.). In addition, DFRE agent 404 can leverage the stored knowledge 414 in DFRE KB 416 to make assessments/inferences.
For example, DFRE agent 404 may perform semantic graph decomposition on DFRE KB 416 (e.g., a knowledge graph), so as to compute a graph from the knowledge graph of KB 416 that addresses a particular problem. DFRE agent 404 may also perform post-processing on DFRE KB 416, such as performing graph cleanup, applying deterministic rules and logic to the graph, and the like. DFRE agent 404 may further employ a definition of done, to check goals and collect answers using DFRE KB 416.
In general, DFRE KB 416 may comprise any or all of the following:
The Knowledge graph also allows different reasoners to:
In other words, DFRE KB 416 acts as a dynamic and generic memory structure. In some embodiments, DFRE KB 416 may also allow different reasoners to share or coalesce knowledge, have their own internal sub-graphs, and/or work collaboratively in a distributed manner. For example, a first DFRE agent 404 may perform reasoning on a first sub-graph, a second DFRE agent 404 may perform reasoning on a second sub-graph, etc., to evaluate the health of the network and/or find solutions to any detected problems. To communicate with DFRE agent 404, DFRE KB 416 may include a bidirectional Narsese interface or other interface using another suitable grammar.
In various embodiments, DFRE KB 416 can be visualized on a user interface. For example, Cytoscape, which has its building blocks in bioinformatics and genomics, can be used to implement graph analytics and visualizations.
Said differently, DFRE architecture 400 may include any or all of the following the following components:
More specifically, in some embodiments, DFRE middleware 402 may include any or all of the following:
DFRE middleware 402 may also allow the addition of new services needed by different problem domains.
During execution, DFRE agent 404 may, thus, perform any or all of the following:
To address combinatorial explosion, the DFRE knowledge graph may be partitioned such that each partition is processed by one or more DFRE agents 404, as shown in
As would be appreciated, graph decomposition can be based on any or all of the following:
In further embodiments, the DFRE framework may also support various user interface functions, so as to provide visualizations, actions, etc. to the user. To do so, the framework may leverage Cytoscape, web services, or any other suitable mechanism.
At the core of the techniques herein is a knowledge representation metamodel 700 for different levels of abstraction, as shown in
One key advantage of the DFRE knowledge graph is that human level domain expertise, ontologies, and goals are entered at the L2 level. This leads, by definition, to an unprecedented ability to generalize at the L2 level thus minimizing the manual effort required to ingest domain expertise.
More formally:
In various embodiments, L2 may comprise both expertise and experience stored in long-term memory, as well as a focus of attention (FOA) in short-term memory. In other words, when a problem is triggered at L*, a DFRE agent 702 that operates on L2-L0 may control the FOA so as to focus on different things, in some embodiments.
As would be appreciated, there may be hundreds of thousands or even millions of data points that need to be extracted at L0. The DFRE’s FOA is based on the abstraction and the DFRE knowledge graph (KG) may be used to keep combinatorial explosion under control.
Said differently, metamodel 700 may generally take the form of a knowledge graph in which semantic knowledge is stored regarding a particular system, such as a computer network and its constituent networking devices. By representing the relationships between such real-world entities (e.g., router A, router B, etc.), as well as their more abstract concepts (e.g., a networking router), DFRE agent 702 can make evaluations regarding the particular system at different levels of extraction. Indeed, metamodel 700 may differ from a more traditional knowledge graph through the inclusion of any or all of the following, in various embodiments:
As noted above, voice controls are becoming increasingly ubiquitous across a variety of use cases. Indeed, many personal computing devices now include voice control functionality. In addition, stand-alone voice control devices are also increasing in popularity for use in home automation and as personal voice assistants.
With the proliferation of voice controls also comes increasing threats of malicious use. In a simple case, an unauthorized user may purposely issue a voice command to create harm. In more sophisticated attacks, malware can cause an infected device to issue voice commands to a voice control device or cause an infected voice control device to perform an unwanted action.
The techniques herein propose leveraging the cognitive metamodel herein for purposes of implementing a secure microphone agent. In some aspects, the secure microphone agent may detect voice commands suspected of being malicious and require authorization before performing the corresponding action. In further aspects, the microphone agent may be integrated directly onto the device receiving the voice command.
Illustratively, the techniques described herein may be performed by hardware, software, and/or firmware, such as in accordance with the DFRE process 248, which may include computer executable instructions executed by the processor 220 (or independent processor of interfaces 210), to perform functions relating to the techniques described herein.
Specifically, according to various embodiments, a device extracts a voice command from audio data captured by a microphone. The device uses a semantic reasoning engine, to determine a goal of the voice command. The device determines that the goal of the voice command is consistent with prior voice commands issued to the device. The device raises an alert when the goal of the voice command is inconsistent with prior voice commands issued to the device.
Operationally,
As shown, one or more microphones may capture intercepted audio 802. Typically, the capturing microphone(s) may be integrated as part of the device executing secure microphone agent 800. However, the techniques herein are not limited as such and intercepted audio 802 could also be sent to the executing device for process.
As an initial processing step, the executing device may perform a speech to text operation 804 on intercepted audio 802. In general, speech to text operation entails generating textual words and phrases based on the audio signals in intercepted audio 802. As would be appreciated, any suitable speech to text engine may be used to perform speech to text operation 804.
Once text to speech operation 804 has been performed, secure microphone agent 800 may then execute a Natural Language Understanding (NLU) parser 806 on the resulting text. Generally, NLU parser 806 is responsible for parsing the text of the issued voice command, to separate the different words of the voice command into different categories. For instance, a typical voice command may include a wake word or phrase (e.g., “Hey Alexa,” “Hey Siri,” etc.). After such a wake word or phrase, the voice command will also typically specify an action (e.g., “turn on,” “order,” etc.), a subject (e.g., “the living room light,” “a box of soap,” etc.), and/or other parameters (e.g., “50% brightness,” “100 units,” etc.).
Typically, keyword/wake word detection 808 will also be performed, to allow the system to distinguish between background utterances and voice commands issued specifically to the voice control device. Accordingly, keyword/wake word detection 808 may entail determining whether the parsed text from NLU parser 806 also includes a predefined wake word or phrase for the voice control device.
According to various embodiments, metamodel 700 may be configured to assess the goal of any voice commands signaled by keyword/wake word detection 808. To do so, metamodel 700 may leverage a semantic reasoner 810 that may be built using a knowledge base representing various concepts and their relationships and actions. For instance, the various actions, subjects, and parameters of voice commands may be represented in the knowledge base of semantic reasoner 810, to make inferences about the overall goal of the voice command. For instance, if the voice command takes the form of “Hey Alexa, turn the living room lights to 50% brightness,” reasoner 810 may determine that the overall goal of the voice command is to control the living room lights in a specific way.
In addition to reasoner 810 determining the goal of an issued voice command, metamodel 700 may also perform goal evaluation 812 on that determined goal, to determine whether the voice command is potentially malicious. In various embodiments, metamodel 700 may do so in part by comparing the goal of the voice command to those of previously issued voice commands. For instance, assume that no user has ever issued a voice command to make an online purchase and that the current voice command is to do so. The fact that this is inconsistent with the goals of the prior voice commands could indicate that the current voice command is malicious. In a more specific case, metamodel 700 can also evaluate the parameters of the voice command as part of its goal evaluation 812, as well. For instance, say a user typically orders five bars of soap, but that the current voice command seeks to order one hundred. In such a case, metamodel 700 may determine that this voice command is inconsistent with the prior commands.
In various embodiments, metamodel 700 may also take into account additional context information, during its evaluation of the voice command. For instance, metamodel 700 may also perform speaker identification 814, to determine whether the issuer of the voice command is a known user. In some embodiments, speaker identification 814 may entail generating a voice signature for the voice command (e.g., based on intercepted audio 802) and comparing that signature to a known list of voice signatures. Such a list may include voice signatures for users that have explicitly registered with the voice control device or users that have previously issued voice commands to it. Thus, if the voice command was issued by an unknown user, metamodel 700 may take this into account during its evaluation of the voice command, as well.
In one embodiment, metamodel 700 may also perform human/machine classification 816, which seeks to distinguish between human-issued voice commands and machine-issued voice commands. Indeed, a potential attack vector against voice control devices is have a malware-infected device issue the voice command, to ‘trick’ the voice control device into performing certain actions. This can be done either by synthetically generating a voice command or performing a replay attack by recording a previous voice command and replaying that voice command at a later time. In either case, metamodel 700 may also take into account the behavior, motion, and/or location of the issuer of the voice command, to perform its human/machine classification. For instance, if the voice command is issued from a new location in a room, at an anomalous time (e.g., at 3:30 AM, when all prior voice commands were issued during the day), or the like, metamodel 700 may determine that the voice command was issued by a machine, instead of a human.
Based on its evaluation of the voice command, metamodel 700 may opt to raise an alert regarding the voice command, before taking any further action with respect to that command. Such an alert may, for instance, take the form of a request for authorization that could be sent for review by an administrator (e.g., the homeowner, etc.). If approved, metamodel 700 may register the voice command in its knowledge base of allowed voice commands, thereby learning that its goal, issuer, etc. are considered benign. In addition, metamodel 700 may cause the voice command to be enacted, such as by issuing a command to a device, sending a request to a service, etc. Conversely, if the alert is disapproved, metamodel 700 can also learn that the goal, issuer, etc. of the voice command are suspect and use this acquired knowledge when evaluating future voice commands.
In various embodiments, speech to text operation 804, NLU parser 806, and/or keyword/wake word detection 808 may be performed as part of secure microphone agent 800 or in conjunction therewith. For instance, the techniques herein provide for secure microphone agent 800 to be fully integrated with a voice control device, such as a mobile device, a stand-alone voice control device (e.g., Amazon Echo, Google Nest, etc.), a smart device (e.g., a thermostat, a security system keypad, etc.), or the like. However, secure microphone agent 800 may also be implemented as a plugin to such a device that already includes these functions, in further embodiments.
Assume now that that a malicious actor 902 issues a voice command 904 to device 908. Such a malicious actor 902 may take the form of a human attempting to perform a malicious action or may be a speaker-equipped machine that issues voice command 904 (e.g., through the execution of malware). In either case, malicious actor 902 may have certain characteristics that can be assessed by secure microphone agent 800, such as their voice fingerprint, location, movement, behavior, etc.
As noted previously, device 908 may perform its various processing steps, to first determine that voice command 904 is a voice command (e.g., by performing speech to text operation 804, executing NLU parser 806, etc.). This may be done to distinguish between normal conversation and voice commands issued directly to device 908.
In turn, secure microphone agent 800 may perform its various analysis, such as goal evaluation 812, speaker identification 814, and/or human/machine classification 816, to determine that voice command 904 is not consistent with previously issued voice commands. For instance, say that voice command 904 is to buy ten quantities of a certain product. If the previous voice commands were for a much lower amount of that product, secure microphone agent 800 may determine that voice command 904 is inconsistent.
Of course, secure microphone agent 800 may also take into account other factors, as well, when evaluating voice command 904. For instance, if secure microphone agent 800 determines that malicious actor 902 does not have a known voice signature, is a machine, has an anomalous location or behavior, or the like, secure microphone agent 800 may likewise deem voice command 904 as an inconsistent voice command.
Based on its determination that voice command 904 is inconsistent, secure microphone agent 800 may raise one or more alerts. In some instances, such an alert may take the form of an audio alert 918 presented via speakers 916 of device 908. In further instances, such an alert may be provided as a visual alert to a display of device 908.
As shown, in a further embodiment, secure microphone agent 800 may send alert 910 to a remote device 912 associated with an administrative user 914 that has previously registered with the voice control functions of device 908. Alert 910 may, for instance, indicate the conclusion/inference made by secure microphone agent 800 regarding voice command 904. In addition, alert 910 may also allow administrative user 914 to allow or deny voice command 904.
If administrative user 914 authorizes voice command 904, secure microphone agent 800 may permit device 908 to enact the voice command (e.g., by making the desired purchase). However, if administrative user 914 denies voice command 904, secure microphone agent 800 may take actions such as preventing voice command 904 from being enacted, disabling microphone 906 on device 908, or the like.
At step 1115, as detailed above, the device may use a semantic reasoning engine to determine a goal of the voice command. In some embodiments, the semantic reasoning engine may do so using a knowledge base representing different concepts and their relationships. For instance, the semantic reasoning engine may determine that the goal of the voice command is to make a certain purchase, control a certain device, etc.
At step 1120, the device may determine that the goal of the voice command is consistent with prior voice commands issued to the device, as described in greater detail above. For instance, the device may determine that an amount of a good being ordered via the voice command is inconsistent with previous orders. Indeed, if the goal of the voice command is not within the range of previous voice commands, this may be an indication of a voice-based attack. In some embodiments, the device may make this determination based in part by determining that a voice fingerprint of the voice command does not match any voice fingerprints of any users that issued the prior voice commands to the device. In another embodiment, the device may make this determination based on a location at which the voice command was issued.
At step 1125, as detailed above, the device may raise an alert when the goal of the voice command is inconsistent with prior voice commands issued to the device. In some embodiments, the device may receive an authorization for the voice command, after raising the alert. For instance, the device may send the alert to a display locally, or send the alert to a mobile device operated by an administrative user. If that user authorizes the voice command, the device may enact the voice command, based on the authorization. For instance, the device may send instructions to a particular device, service, or the like, to perform the voice command. Procedure 1100 then ends at step 1130.
It should be noted that while certain steps within procedure 1100 may be optional as described above, the steps shown in
While there have been shown and described illustrative embodiments that provide for a secure microphone agent using semantic reasoning, it is to be understood that various other adaptations and modifications may be made within the spirit and scope of the embodiments herein. For example, while certain embodiments are described herein with respect to specific types of sensor systems, the techniques can be extended without undue experimentation to other use cases, as well.
The foregoing description has been directed to specific embodiments. It will be apparent, however, that other variations and modifications may be made to the described embodiments, with the attainment of some or all of their advantages. For instance, it is expressly contemplated that the components and/or elements described herein can be implemented as software being stored on a tangible (non-transitory) computer-readable medium (e.g., disks/CDs/RAM/EEPROM/etc.) having program instructions executing on a computer, hardware, firmware, or a combination thereof. Accordingly, this description is to be taken only by way of example and not to otherwise limit the scope of the embodiments herein. Therefore, it is the object of the appended claims to cover all such variations and modifications as come within the true spirit and scope of the embodiments herein.
