Multiplex communications – Data flow congestion prevention or control
Reexamination Certificate
1999-07-28
2003-06-03
Patel, Ajit (Department: 2664)
Multiplex communications
Data flow congestion prevention or control
C370S395640, C370S412000
Reexamination Certificate
active
06574193
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates generally to communication networks, and specifically to transmission of audio data in ATM networks.
BACKGROUND OF THE INVENTION
Asynchronous Transfer Mode (ATM) has been widely accepted as a packet switching communications standard for high-speed, Broadband Integrated Services Digital Networks (B-ISDN). ATM networks are designed to carry a mixture of data, voice and video on common circuits. An ATM packet, or cell, comprises 53 bytes, including a five-byte header and 48-byte payload.
The International Telecommunications Union (ITU) has promulgated standard I.363.2, “B-ISDN ATM Adaptation Layer Type 2 Specification” (September, 1997), known as AAL-2, in order to enable efficient transmission of real-time voice-band data over ATM networks. The standard, which is incorporated herein by reference, defines how traffic in the form of packets of various numbers of octets (or bytes) are assembled into ATM cell payloads at a transmitter and disassembled back into the original packet stream at the receiver.
ITU standard I.366.2, “AAL Type 2 Service Specific Convergence Sublaver for Trunking,” which is incorporated herein by reference, defines several types of packet formats for voice-band data, which are encoded into packets of variable length. A Type-1 packet, as defined by this standard, has no explicit identifier for the encoding algorithm used. The encoding algorithm is implicitly identified, however, by a combination of the packet length, a parameter in the packet header (a user-user interface code, or UUI), and a predefined profile agreed upon between the transmitter and the receiver. The standard identifies the profile as “a mapping that informs the receiver of a Type-1 audio packet how to interpret the packet content. The domain of this mapping is a set of pairs (UUI, Length).” The first element of each pair is a UUI codepoint in the range 0-15, which is utilized in part to indicate a packet sequence number. The second element is the packet length for one of the encoding formats included in the profile. Each Type-1 packet may include multiple encoded packets, or sub-packets.
Each encoding algorithm produces some minimum encoded packet length, dependent on the encoding bit rate. For example, packets encoded using the LD-CELP algorithm, in accordance with standard G.728 of the ITU, have minimum packet lengths of 10, 8 and 6 octets, respectively, for bit rates of 16, 12.8 and 9.6 Kbit/s, at which the LD-CELP algorithm is conventionally allowed to work. A profile may also include a combination of encoded packets with more than one minimum length interval, if no ambiguity occurs as a result. Thus, for example, the profile may include LD-CEL encoded packets of lengths 20, 16 and 12 octets. Part of the UUI field may be used to expand the profile if ambiguity does occur. The cost of expanding the number of bits allocated to the profile is that the modulus of the packet sequence number, which is used to detect irregularities in the packet stream at the receiving end, is reduced.
The alternative packet lengths and corresponding bit rates of a given encoding scheme may be used for controlling traffic rate at the input to an ATM network from a transmission source. This technique is known generally as variable rate encoding (VRE). The ITU defines VRE for use in voice transmission over ATM network as “the capability of the encoding algorithm to dynamically switch between the nominal bit rate to lower or higher discrete bit rates under the control of the ATM equipment.”
How VRE is Performed depends on the signal type, for example, speech, fax or modem, and the kind of encoding algorithm used, such as the ADPCM, LD-CELP or CS-ACELP algorithms, for example, which are defined by the ITU G. 726, G.728 and G.729 standards, respectively. For each particular signal type and algorithm, there is a set of applicable, discrete output bit rates, such as the LD-CELP rates of 9.6, 12.8 and 15 Kbit/s mentioned above. A VBD-modem signal, as defined by the G.726 standard, is commonly encoded at 40 Kbit/s. Typically, VRE on speech signals is performed by selecting among the applicable bit rates. Another option, useful for fax signals with forward error correction (FEC), for example, is to change the bit rate by using or not using additional FEC bits.
The use of VRE is generally required in the following situations:
When the traffic load in a specific ATM Virtual Circuit (VC) is higher than the bandwidth allowed by contract with the network being used; or
When the sum of the traffic loads in all of the VC's that share a common link exceeds the link bandwidth.
Both of these conditions may occur simultaneously. There is also the possibility that traffic load on a given Virtual Path (VP, as defined by the ITU and ATM Forum), which serves a number of VC's, mav exceed a contractual bandwidth.
U.S. Pat. No. 5,646,943, to Elwalid, which is incorporated herein by reference, describes a congestion control method in which an access regulator limits the rate of input data to a network node. A rate parameter of the access regulator is dynamically adjusted according to a level of congestion in the node.
U.S. Pat. No. 5,805,577, to Jain, et al., which is incorporated herein by reference, similarly describes a congestion avoidance scheme for data traffic in ATM networks. Input data rates from a number of active sources transmitting data via a network switch are measured and used to determine an overload factor relative to available bit rate capacity. The overload factor is used to dynamically determine a fair share allocation of transmission capacity among the sources. The transmission rates of the sources are adjusted so as not to exceed the respectively allocated capacity. This scheme is complex and computationally heavy, since it requires that the input data rates be measured, and that all of the queues associated with the data sources be viewed and taken into account.
U.S. Pat. No. 5,463,620, to Sriram, and an article by Sriram entitled “Methodologies for Bandwidth Allocation, Transmission Scheduling, and Congestion Avoidance in Broadband ATM Networks,” in
Computer Networks and ISDN Systems
26 (1993), pp. 43-59, which are incorporated herein by reference, describe a congestion control mechanism using bit-dropping of packetized voice data. When congestion occurs, traffic loads is reduced by dropping the least-significant bits of already-encoded voice packets.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide an improved method for congestion control in ATM networks.
It is a further object of some aspects of the present invention to provide a method for congestion control that is particularly applicable to voice-band data utilizing the AAL-2 standard.
In preferred embodiments of the present invention, one or more variable-rate encoders generate encoded data for transmission via a virtual circuit (VC) over an ATM network. A degree of circuit congestion, or stress, is determined, based on a level of congestion of the network encountered by the VC. The stress level is fed back to the encoders, which vary their encoding rates responsive thereto, so as to adjust their output bit rate to the congestion level.
In some preferred embodiments of the present invention, the encoded data are assembled into cells, preferably using an AAL-2 service, as is known in the ATM art. The cells enter a cell queue for transmission via the VC. A fill level of the cell queue, i.e., an expected length of the queuing delay, is used to determine the stress level for controlling the encoding rates of the encoders.
In some preferred embodiments of the present invention, a plurality of VC's, each with its own cell queue, are multiplexed over a network link. The stress level for each VC, as well as the corresponding encoding rates of the one or more encoders associated with the VC, is determined substantially independently for each of rune VC's. Mutual influence among the VC stress levels is felt indirectly, however, when network bandwidth limitations cau
Abelman ,Frayne & Schwab
Patel Ajit
Shah Chirag
Veraz Networks Ltd.
LandOfFree
Congestion control using variable rate encoding based on... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Congestion control using variable rate encoding based on..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Congestion control using variable rate encoding based on... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3099245