The invention is related to the field of telecommunication devices and services and more specifically, the invention is directed to an apparatus and method that prioritizes and improves the flow of voice traffic in certain portions of a data network that also functions, in part, as a telecommunication system utilizing Voice over Internet Protocol (VoIP).
VoIP is a technological development in the field of telecommunications that is utilized to transmit voice conversations over a data network using Internet Protocol (IP) packets rather than the existing and traditional telecommunications system more commonly referred to as the Public Switched Telephone Network (PSTN) or Plain Old Telephone Service (POTS). Entities (e.g., businesses or individuals) implement VoIP by purchasing and installing the necessary equipment (e.g., one or more Customer Premise Equipment (CPE) devices) to access a VoIP service provider and activating this telecommunication service via a broadband Internet connection.
Typically, entities will not have solely telecommunications equipment connected to the broadband Internet connection. Various other data networking devices including but not limited to computers, peripherals and wireless networking devices will comprise a substantial Local Area Network (LAN) that is connected to a Wide Area Network (WAN) with a multitude of services available via the broadband Internet connection. With the growing amount video and rich content (iTunes, You-Tube etc.) being available on the Internet, entities are increasingly downloading large files, which has the potential to saturate a broadband Internet connection in the downstream direction (i.e., towards the LAN). P2P applications like Bit-Torrent are also starting to inundate broadband Internet connections in both directions. Some VoIP equipment have upstream Quality of Service (QoS) capabilities which prioritize voice-type traffic ahead of other application traffic, thereby ensuring high voice quality towards the remote party (towards the WAN). However, QoS capabilities are limited in the downstream direction, which causes VoIP customers to experience poor voice quality when someone on their LAN (i.e., a home network) starts a large download.
The downstream bandwidth between the server 116 and a CMTS 112 is typically higher than that of the customer's broadband link. Therefore, when the customer starts a download 123, typically with a Transmission Control Protocol/Internet Protocol (TCP/IP)-based protocol, TCP rate control quickly ratchets up the downstream bandwidth utilization to the level of the lowest bandwidth link in the chain, which is usually the customer's broadband link. If, during this download, a voice communication session 121 occurs, there is packet queuing (delay & jitter) and packet drops 113 at the CMTS 112 in the downstream direction, both of which contribute to poor voice quality. The upstream bandwidth is usually not severely affected, since TCP ACK packets in the upstream direction are of very small size (typically less than 50 bytes), and fewer in number than downstream TCP packets.
Accordingly, there is a need in the art for an apparatus and method for providing voice traffic management in a data network to improve the overall experience of VoIP telephony.
The disadvantages associated with the prior art are overcome by a method and apparatus of voice traffic management in a data network that functions in part as a telecommunication system based on VoIP. The data network has a wide area network (WAN) portion and a local area network (LAN) portion. In one embodiment, the method includes the steps of establishing a default maximum bandwidth setting at a LAN egress port when voice-type traffic is not present in the LAN portion of the data network, detecting voice-type traffic in the LAN portion of the data network, reducing the bandwidth setting at the LAN egress port to effect a change in a rate of non voice type traffic entering the LAN portion of the data network and monitoring non voice type traffic and voice quality statistics to determine if the rate of non voice type traffic entering the LAN portion of the data network has changed. Once the desired change has occurred, the method continues by performing a linear increase of the bandwidth setting at the LAN egress port to a first value while monitoring voice quality statistics, determining if voice quality has degraded during increase of the bandwidth setting and repeating the last two steps if voice quality has not degraded. Preferably, the bandwidth setting at the LAN egress port is set to a second value that is less than the first value.
The method further includes the steps of upon detection of termination of the voice-type traffic in the LAN portion of the data network, setting the bandwidth setting to the default maximum value and upon detection of new voice-type traffic in the LAN portion of the data network, setting the bandwidth setting to a fraction of the second value. Preferably, the fraction of the second value is approximately 70%. The maximum value is determined by one selected from the group consisting of a measurement of the downstream bandwidth of the broadband link associated with the data network and a sum of the second value and the bandwidth used by the voice-type traffic. The detection of voice-type traffic in the LAN portion of the data network is achieved by one selected from the group consisting of determining completion of SIP/SDP codec negotiation and determining the sending or receiving of a SIP INVITE message associated with the voice-type traffic in the LAN portion of the data network. The bandwidth setting is only reduced with respect to non-voice type traffic in the LAN portion of the data network that effects voice quality. Further, parts of the presented method can be performed two or more times consecutively to account for two or more instances of new voice-type traffic in the LAN portion of the data network.
So that the manner in which the above recited features of the present invention are attained and can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments thereof which are illustrated in the appended drawings.
It is to be noted, however, that the appended drawings illustrate only typical embodiments of this invention and are therefore not to be considered limiting of its scope, for the invention may admit to other equally effective embodiments.
To facilitate understanding, identical reference numerals have been used, where possible, to designate identical elements that are common to the figures.
To achieve the desired objectives, the subject invention provides IP-based telecommunication services (i.e., VoIP) subscribers with the ability to dynamically control the bandwidth of data entering a LAN portion of a larger data network to improve voice communications therein. Real-time management and control are executed by an apparatus and method in an IP-based environment (i.e., the Internet). In a preferred embodiment of the invention, the voice traffic management functions operate as part of a larger VoIP telecommunication service. Such telecommunication service is, by way of example, part of any public or private data network (or combination thereof) constructed for (in part) and adapted to convert analog voice signals (e.g., generated by a human utterance) to a digitized and packetized format according to known and understood protocols (such as but not limited to Transmission Control Protocol/Internet Protocol (TCP/IP)) for transmission from an originating point (Party A) to one or more terminating points (Party B and/or C, D and the like). In a preferred embodiment of the invention, the data network is an IP-based network such as (but not limited to) the Internet having VoIP specific and related components connected thereto.
At step 206, new voice-type traffic is detected in the data network. Voice-type traffic is further defined by, but not necessarily limited to, signaling information used to establish a voice communication session and the actual media stream that carries the content of the voice communication. In one embodiment of the invention, an example of the signaling information is selected from the group consisting of Session Initiation Protocol (SIP) messages and Session Description Protocol (SDP) parameters. SIP is a packed network based communications protocol that is used to establish, tear down and provide additional features and functionality to VoIP telephony. The details and functionality of SIP can be found in the Internet Engineering Task Force (IETF) Request for Comments Paper No. 3261 herein incorporated in its entirety by reference. SDP is a format for describing streaming media initialization parameters in an ASCII string. The details and functionality of SDP can be found in the IETF Proposed Standard as RFC 4566 herein incorporated in its entirety by reference. In one embodiment of the invention, an example of the media stream is the Real-time Transport Protocol (RTP) which defines a standardized packet format for delivering audio and video over the Internet. RTP is frequently used in streaming media systems as well as in videoconferencing and push to talk systems. For these it carries media streams controlled by H.323 or SIP signaling protocols, making it the technical foundation of the Voice over IP industry. The details and functionality of RTP can be found in IETF RFC 3550 herein incorporated in its entirety by reference. In one embodiment of the invention, detection of voice-type traffic is determined by completion of SIP/SDP codec negotiation, but other detection triggers are possible to those skilled in the art of VoIP telephony.
At step 208, the bandwidth at the LAN egress port associated with a LAN portion of a data network is significantly reduced from its previously established default value. The purpose of this bandwidth reduction is to affect the rate of non-voice data that is coextensively traveling from the WAN portion of the data network to the LAN portion of the data network when a new voice communication is being attempted. Preferably and in one embodiment of the invention, the reduced bandwidth is approximately 150 kbps. This reduced bandwidth value however is configurable and one skilled in the art can arrive at a suitable value alternate values which are also within the scope of the invention to achieve the necessary bandwidth reduction. This reduction in bandwidth severely throttles user application traffic that is not part of the voice communication. For example, scheduling of network traffic in the LAN 101 is implemented by a weighted fair queuing (WFQ) algorithm between a plurality of queues as described in greater detail below with respect to
In one embodiment of the invention, such reduction occurs subsequent to completion of SIP/SDP codec negotiation but prior to establishment of the RTP session that characterizes the new voice communication. This allows the rate limiting to take effect a few hundred ms earlier, thereby reducing the effect of the downstream traffic on the voice quality at the beginning of the call. In an alternate embodiment of the invention, the bandwidth reduction occurs immediately upon sending or receiving a SIP INVITE, which provides additional time for the method to achieve convergence. In another alternate embodiment of the invention, the bandwidth reduction occurs in a selective rather than universal manner. That is, in the first two embodiments, the bandwidth is reduced without regard to the specific downstream data traffic that is entering the LAN portion of the data network (i.e., any number of file downloads, web pages and other activity). In the selective alternate embodiment, the active download stream(s) that is(are) responsible for reducing voice quality are identified and limited thereby leaving other low-usage open TCP streams alone. This will improve the user experience for the other applications, while only affecting the offending applications. For example, the active download stream(s) that have a significant impact on voice quality are identified by tracking the state of each active TCP session, and maintaining downstream bandwidth utilization statistics for each TCP session. The session(s) that have high bandwidth utilizations will have the most impact on downstream voice quality, and these are selectively limited while leaving those session(s) with low bandwidth utilizations unimpacted.
At step 210, a decision is executed as to whether non-voice traffic has dropped below a desired threshold. That is, non-voice traffic queuing is monitored to determine if the reduction in bandwidth at the LAN egress port has had any effect. The amount of non-voice traffic is compared to a Queue Threshold Factor (QTF, a previously established value). If the non-voice traffic is not below the QTF, the method 200 proceeds to step 212 where a waiting period is invoked or otherwise determined to be in effect to allow the non-voice traffic an opportunity to fall below the QTF. The method 200 then loops back to repeat step 210 until non-voice traffic is at or below the QTF.
At step 214, a decision is executed as to whether the quality of the voice communication session has degraded as a result of the bandwidth reduction of step 208 and non-voice traffic volume. If the voice communication session quality has degraded, the method 200 proceeds to step 216 where the bandwidth is further reduced to a fraction of the current value (as set forth in step 208). The method 200 then loops back to repeat step 214 until voice communication session quality has not degraded wherein, the method 200 proceeds to step 218.
At step 218, the bandwidth at the LAN egress port is increased as voice communication quality statistics are monitored. More specifically, once the TCP rate of the non-voice traffic in the data network is scaled back, the telephone adapter allows for an increase in non-voice traffic entering the LAN portion of the data network. Preferably, this increase is linear in profile with a slope characteristic that is configurable. In one example of the invention, the slope is configured to be 10 kbps per second. The bandwidth is increased until a first value is reached. Such first value is defined and explained in greater detail below with respect to a LAN Egress Manager Module 632 in one embodiment of the invention. While the increase is being performed, RTCP statistics for the incoming stream(s) (i.e., voice communication session(s)) are monitored.
At step 220, a decision is executed as to whether the quality of the voice communication session has degraded as a result of the increase in bandwidth associated with step 218. Specifically, if the RTCP statistics indicate an increase in packet drops, jitter (packet discards), delay or MOS score, then the method proceeds to step 222 to reduce the bandwidth at the LAN egress port. In one embodiment of the invention, the reduction occurs in an exponential manner to a fraction of its current value, though other response profiles are possible and known to those skilled in the art. The method then once again proceeds to step 218 to linearly increase the bandwidth at the LAN egress port. However, the bandwidth is now increased only up to a value that is less than the first value (where RTCP statistics indicated a previous drop in voice quality) and the method re-enters step 220 to perform voice communication quality analysis.
If no degradation in voice quality is determined by the decision of step 220, then the proceeds to step 224 wherein a second value is determined. This second value is determined and then defined as the bandwidth “equilibrium” value. Such second value is stored in memory for future voice-type traffic processing scenarios as described in greater detail below.
When the voice communication ends at step 226, the default maximum bandwidth is re-established as it is approximately the maximum value of the downstream bandwidth of the broadband link. Performing this step allows the LAN portion of the network to receive data as it normally would, unimpeded by voice-type traffic override protocols as described in the subject method. This maximum value is either determined through downstream bandwidth measurement, or estimated based on the equilibrium value and the bandwidth used by the voice-type traffic.
At step 228, the bandwidth at the LAN egress port is once again adjusted upon the detection of one or more subsequent voice communication sessions. In a preferred embodiment of the invention, when the downstream RTP session is established for the subsequent voice communication session(s), the bandwidth at the LAN egress port is adjusted to a value which is a fraction of the equilibrium/second value. This fraction is fine tuned to allow the telephone adapter to quickly converge to the equilibrium value. In one embodiment of the invention, the fraction is a configurable value and preferably set at 70% The method ends at step 230.
In this first embodiment of the invention, an improved telephone adapter (TA) 302 described in greater detail below replaces the prior TA 104. The voice traffic management method 200 described above is executed within the improved TA 302 such that the desired bandwidth control of non-voice-type traffic is achieved. Specifically, upon detection of a new voice communication, the improved TA 302 alters bandwidth settings at its LAN egress port such that only non-voice-type traffic 304 is queued and potentially dropped at the improved TA 302 rather than proceeding into the LAN portion 101 of the data network 300. Non-voice-type traffic 304 includes but is not limited to user application packet attempting to move through the LAN portion 101 as part of a download 123 from server 116 to a laptop device 110 on the customer network 105. By controlling the downstream rate to the laptop device 110, at the LAN egress port, non-voice-traffic is queued (and possibly dropped). This causes the TCP rate control to scale down the download rate to a point where packet drops are avoided at the CMTS 112.
The router 608 is connected to a classifier 612 which identifies and orders (or queues) incoming data packets for distribution to the appropriate device(s) in the LAN 101. The classifier 612 includes a dynamic class which exists by default, called VoIP Service Provider Class that identifies downstream voice communication sessions based on header fields selected from the group consisting of: Source IP Address, Destination IP Address, Source IP Port, Destination IP Port, Protocol and TOS/DS-Byte field. In one embodiment of the invention, the classifier 612 supports a configurable number of entries for this class in order to support multiple voice communication sessions. These entries are automatically populated by the snooper function described below.
The classifier 612 is connected to a plurality of queues 614 which retain the data packets as organized by the classifier 612 for subsequent output to a LAN connected device. The plurality of queues 614 is connected to a scheduler 616 which maintains the chronological order of the data packets outputted from the plurality of queues 614. The scheduler 616 is connected to a traffic shaper 618 that adjusts the rate of egress traffic towards the LAN 110 to a specific bandwidth dependent upon traffic conditions. The bandwidth is controlled by the traffic control manager 610 as described below. The traffic shaper 618 is connected to a LAN driver 620 for facilitating IP stack management and data packet movement out of the TA 600. LAN driver 620 is connected to LAN port hardware 622 adapted for receiving data network traffic conveyance media (i.e., RJ-45 cabling and the like) to define a LAN egress port.
The router 608 is also connected to the traffic control manager 610 for the purposes of evaluating the type of traffic (e.g. voice-type and non-voice-type) moving through the TA 600 so that the appropriate traffic management actions can be taken in accordance with the subject invention such as but not limited to evaluating the quality of the voice-type traffic. Specifically, the traffic control manager 610 includes a voice-type traffic snooper 630 connected to the router 608. The snooper 630 evaluates voice-type traffic flowing between the LAN port 622 and WAN port 602 to identify which packets belong to VoIP provider voice communication session(s). These include SIP, RTP, RTCP and VoIP provider DNS packets. In one embodiment of the invention, examination of the SDP fields determines which RTP sessions belong to VoIP provider voice communication session(s). The snooper 630 controls the addition (when a new VoIP provider voice communication session is identified) deletion (when an existing VoIP provider voice communication session is closed) of entries in the classifier 612 and assigns them to a VoIP service provider class so that such sessions are not limited by the actions and characteristics of the queues 614, scheduler 616 and traffic shaper 618.
Traffic control manager 610 further includes a plurality of LAN egress control modules that assist in executing the bandwidth monitoring and control functionality described in the method above. Specifically, a LAN egress manager module 632 dynamically controls the shaper's bandwidth. The LAN egress manager module 632 is connected to a LAN egress manager statistics module 634 which maintains network traffic statistics, a LAN egress manager logs module 636 which maintains logs based on the collected statistics, a LAN egress manager configuration module 640 and an RTCP Statistics module 638.
The LAN Egress Manager 632 continuously monitors the queues 614, which would initially fill up when the shaper is enabled 618, and slowly drop down when the TCP rate control kicks in. When an average queue utilization drops below a value defined by a Queue Threshold Factor, the LAN Egress Manager 632 checks to see if a Shaper Ramp-up Timer has expired, and if a packet Fraction Lost, Fraction Discarded, Round-trip Delay and Listening MOS are all below their respective thresholds. If so, the shaper 618 is ramped up by adding a Linear Shaper Ramp-up Bandwidth Adder value to the current shaper bandwidth value. The LAN Egress Manager 632 waits for the Shaper Ramp-up Timer to expire again, and then checks the packet Fraction Lost, Fraction Discarded, Round-trip Delay and Listening MOS. The calculation is done using cumulative metrics taken from when the last ramp-up occured. If all of them are meet their respective thresholds, it ramps up the shaper bandwidth value again. This process continues until one of the above goes above its threshold. The shaper bandwidth value at that point is stored as the Shaper Maximum Bandwidth (i.e, the first value described above with respect to the method 200).
Once the packet Fraction Lost, Fraction Discarded, Round-trip Delay or Listening MOS goes above its threshold, indicating loss of voice quality, the shaper 618 is ramped down to a value given by (current shaper value*Exponential Shaper Ramp-down Factor). After the Shaper Ramp-down Timer expires, the voice quality metrics are checked again. If any of them are still above their respective thresholds, the ramp-down process is repeated.
Once the voice quality metrics are all below their thresholds, the ramp-up process starts again, but this time it ramps only up to a value given by (Shaper Maximum Bandwidth-Shaper Equilibrium Differential). This value is now re-designated as the new Shaper Maximum Bandwidth (i.e., the second value with respect to the method 200 described above) for the next ramp-down/ramp-up cycle. If none of the voice quality thresholds are crossed when the shaper 618 is at the new Shaper Maximum Bandwidth, then it has converged (reached equilibrium). The Shaper Maximum Bandwidth at the end of the call is stored in memory as Previous Call Shaper Maximum Bandwidth if the equilibrium was achieved in the call, else this value is not updated. When all calls to the device end, the shaper 618 is set to the broadband link bandwidth as described in point earlier. For subsequent voice calls, at the beginning of the call the shaper 618 is started at a value given by (Previous Call Shaper Maximum Bandwidth*Exponential Shaper Ramp-down Factor) or Initial Shaper Bandwidth, whichever is larger.
The memory 704 is coupled to the CPU 702. The memory 704, or computer-readable medium, may be one or more of readily available memory such as random access memory (RAM), read only memory (ROM), floppy disk, hard disk, flash memory or any other form of digital storage, local or remote. The support circuits 706 are coupled to the CPU 702 for supporting the processor in a conventional manner. These circuits include cache, power supplies, clock circuits, input/output circuitry and subsystems, and the like. A software routine 712, when executed by the CPU 702, causes the controller 700 to perform processes of the present invention and is generally stored in the memory 704. The software routine 712 may also be stored and/or executed by a second CPU (not shown) that is remotely located from the hardware being controlled by the CPU 702.
The software routine 712 is executed when the disclosed method of processing group call requests is desired. The software routine 712, when executed by the CPU 702, transforms the general purpose computer into a specific purpose computer (controller) 700 that controls the various servers TA components/modules in the manner described above to achieve the network traffic management desired. Although the process of the present invention is discussed as being implemented as a software routine, some of the method steps that are disclosed therein may be performed in hardware as well as by the software controller. As such, the invention may be implemented in software as executed upon a computer system, in hardware as an application specific integrated circuit or other type of hardware implementation, or a combination of software and hardware. The software routine 712 of the present invention is capable of being executed on computer operating systems including but not limited to Microsoft Windows 98, Microsoft Windows XP, Apple OS X, eCOS and LINUX. Similarly, the software routine 712 of the present invention is capable of being performed using CPU architectures including but not limited to Apple Power PC, Intel x86, Sun service provider agentRC, MIPS and Intel ARM.
Note that while the embodiments described above are with respect to the monitoring of bandwidth to improve voice communications sessions over data network, these are not the only types of sessions that can benefit from the present invention. In general, any other real time communication (e.g. video) that can benefit from downstream QoS management are contemplated in the subject invention. The enhanced TA 302/402 as described can support selective bandwidth limitation in accordance with the invention for media streams other than VoIP? For example, instead of “snooping” for SIP and RTP protocols to identify voice vs. non-voice traffic, it can snoop for the protocol corresponding to the media stream in question (for example RTSP), and apply selective limiting to all other downstream traffic. This snoop based differentiation can apply to multiple such media streams (similar to multiple voice streams presented earlier). The multiple streams (whether they be voice, video or other desired downstream activity) could take several forms, including: a second (or third/fourth/etc) independent calls from either TA 302 or 402, a second leg of an existing call (a 3-way call), putting an existing call on hold (which generally drops the voice stream in one or both directions), or even a change in the voice media type (and therefore bandwidth utilization) of an existing call. Most such multiple streams will be preceded by some SIP signaling, but not always in the case of the voice media type changing.
While foregoing is directed to embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof.
This application is a continuation of co-pending U.S. patent application Ser. No. 12/322,141, titled “METHOD AND APPARATUS FOR VOICE TRAFFIC MANAGEMENT IN A DATA NETWORK,” filed on Jan. 28, 2009, and listing Deepak Ottur as the inventor, which is a non-provisional of U.S. Provisional Patent Application No. 61/062,615 filed Jan. 28, 2008. The Ser. No. 12/322,141 patent application is herein incorporated in its entirety by reference.
Number | Name | Date | Kind |
---|---|---|---|
6967921 | Levy et al. | Nov 2005 | B1 |
7065048 | Levy et al. | Jun 2006 | B1 |
8483210 | Ottur | Jul 2013 | B2 |
20020087711 | Leung | Jul 2002 | A1 |
20030229714 | Kiremidjian et al. | Dec 2003 | A1 |
20050018651 | Yan | Jan 2005 | A1 |
20070025249 | Yeom | Feb 2007 | A1 |
20070061433 | Reynolds et al. | Mar 2007 | A1 |
20070147247 | Kalonji et al. | Jun 2007 | A1 |
20070201446 | Wright | Aug 2007 | A1 |
20080144660 | Godlewski | Jun 2008 | A1 |
20080165685 | Weiss et al. | Jul 2008 | A1 |
20080225709 | Lange | Sep 2008 | A1 |
20080298237 | Dos Remedios et al. | Dec 2008 | A1 |
20090028161 | Fullarton et al. | Jan 2009 | A1 |
20090029777 | Holt et al. | Jan 2009 | A1 |
20090046706 | Chernow | Feb 2009 | A1 |
20090089849 | Jefremov | Apr 2009 | A1 |
20090141631 | Kim | Jun 2009 | A1 |
Number | Date | Country |
---|---|---|
WO 0048366 | Aug 2000 | WO |
Entry |
---|
Internet Engineering Task Force RFC3261, “SIP: Session Initiation Protocol”, J. Rosenberg et al., Jun. 2002. |
Internet Engineering Task Force RFC3550, “RTP: A Transport Protocol for Real-Time Applications”, H. Schulzrinne et al., Jul. 2003. |
Internet Engineering Task Force RFC4566, “SDP: Session Description Protocol”, M. Handley et al., Jul. 2006. |
Karam, M.J. “Analysis of the Delay and Jitter of Voice Traffic Over the Internet” In: INFOCOM 2001 20th Annual Joint Conference of the IEEE (See PCT Search Report). |
Number | Date | Country | |
---|---|---|---|
20130294444 A1 | Nov 2013 | US |
Number | Date | Country | |
---|---|---|---|
61062615 | Jan 2008 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 12322141 | Jan 2009 | US |
Child | 13937410 | US |