Electrical computers and digital processing systems: multicomput – Computer-to-computer data modifying – Compressing/decompressing
Reexamination Certificate
2000-09-26
2003-03-18
Tokar, Michael (Department: 2819)
Electrical computers and digital processing systems: multicomput
Computer-to-computer data modifying
Compressing/decompressing
C709S216000, C709S217000, C341S060000, C370S393000, C370S465000, C370S474000
Reexamination Certificate
active
06535925
ABSTRACT:
FIELD OF THE INVENTION
The invention relates generally to packet communications and, more particularly, to header compression in packet communications.
BACKGROUND OF THE INVENTION
The term header compression (HC) refers to the art of minimizing the necessary bandwidth for information carried in packet headers on a per hop basis over point-to-point links. Header compression is usually realized by sending static information only initially. Semi-static information is then transferred by sending only the change (delta) from the previous header, and completely random information is sent without compression. Hence, header compression is usually realized with a state machine.
Conventional header compression algorithms are designed basically for narrow band wired channels with a rather small complexity at the receiving decompression side. Also, the complexity at the sending compressing side is kept low to allow efficient implementations in routers where as much computing capacity as possible is needed for the routing. Further, the wired channels for which existing header compression algorithms are designed typically have very small probabilities for bit errors (e.g., a bit error rate of 10
−6
). Wireless channels (generally characterized by lossy, narrow bandwidth links) typically have a much higher probability for error, so header compression for use in wireless channels should be designed with a much larger bit error probability in mind (e.g., bit error rates up to 10
−3
).
Conventional compression schemes for RTP/UDP/IP headers are often based on soft-state machines with states called contexts. The de-compressor context is often updated by each packet received, and if a packet is lost on the link, the context will become invalid. When the decompressor context is invalidated, all successive packets have to be discarded until the soft-state is updated by a full (uncompressed) header. A request for update is sent from the receiving end when the decompressor realizes that the first packet is discarded (or lost), and then it takes a full round-trip (from receiving end to transmitting end and back) before the update (packet with uncompressed header) arrives. This often results in many lost packets. The loss of context state may also occur if the receiving de-compressor fails to successfully de-compress a compressed header.
If the payload for the packets with the compressed headers carries a real time service, the loss of several successive packets may be disastrous for the quality of that real time service. For example, the quality of real time speech service will degrade substantially with increased packet loss rate due to successive lost speech frames. If the speech frame errors have a bursty characteristic, the speech quality will be worse than for the same speech frame error rate but with a less correlated frame error characteristic.
One way of reducing the probability for invalid context states, and thereby packet loss, is to increase the intelligence at the receiver, for example by increasing the probability for the de-compressor to successfully estimate (guess) what the correct context state should be, without using more bits per compressed header. In the example of real time speech service, the conventional RTP time stamp field value typically increments in a predictable fashion (and thus can be reliably predicted or guessed) during periods of speech, but after silent or non-speech periods the time stamp has a more randomized value from the receiver's point of view.
The existing standard for compression of RTP/UDP/IP headers (see, e.g., Steven Casner and Van Jacobson,
Compressing IP/UDP/RTP Headers for Low
-
Speech Serial Links,
IETF RFC 2508, IETF Network Working Group, February 1999, incorporated herein by reference) is referred to herein as CRTP. In CRTP the time stamp delta value is coded with a varying number of bits depending on the value. A large time stamp change since the last packet causes a large delta value, which disadvantageously requires more bits in the compressed header to carry the delta value indicative of the time stamp information.
Whenever DTX (discontinuous transmission) or silent suppression is used in a real time speech service, the time stamp field of the RTP header will have a stochastic behavior difficult to predict in a stream of RTP/UDP/IP packets carrying speech. Hence, the time stamp field is one of the most difficult fields to de-compress at the receiver by means of guessing. In CRTP, the time stamp delta value is coded with a number of bits that depends on the size of the time stamp change since the last packet. Thus, long silent or non-speech periods require more bits to delta-modulate the time stamp field, so the first header after a silent period will typically be larger than in speech packets corresponding to a speech period.
It is therefore desirable to provide a technique for header compression/decompression without the aforementioned disadvantages associated with conventional schemes.
SUMMARY OF THE INVENTION
The present invention advantageously provides techniques for efficiently compressing and reconstructing the time stamp value of a real time communications packet whose time stamp value does not fall within a normally expected sequence of time stamp values. A first part of the time stamp value is selected by the header compressor and transmitted. A second part of the time stamp value is estimated by the header decompressor based on elapsed time between receipt of consecutive packets. The header decompressor combines the second part with the first part received from the header compressor to produce a reconstructed time stamp value.
Further, the present invention is more particularly directed to providing techniques for efficiently compressing and reconstructing a header of a real time communication packet. In a header compressor, a header field value (which may be scaled) is applied to a modulo X operator which can, for example, divide the header field value by a value X, and output the remainder. Optionally, a checksum may be appended to the remainder. Thus, the compressed header field includes the remainder as output from the modulo X operator, with or without the checksum appended therto. A header decompressor includes a field reconstructor which reconstructs the received compressed header field in response to the remainder value and range information. The range information represents a range of possible field values which can be reconstructed from the received remainder value.
A more complete appreciation of the present invention and the scope thereof can be obtained from the accompanying drawings which are briefly summarized below, the following detailed description of the presently-preferred embodiments of the invention, and the appended claims.
REFERENCES:
patent: 5144623 (1992-09-01), Bernardini
patent: 5293379 (1994-03-01), Carr
patent: 5345451 (1994-09-01), Uriu et al.
patent: 5521940 (1996-05-01), Lane et al.
patent: 5813011 (1998-09-01), Yoshida et al.
patent: 5835730 (1998-11-01), Grossman et al.
patent: 6032197 (2000-02-01), Birdwell et al.
patent: 6292840 (2001-09-01), Blomfield-Brown et al.
patent: 6317433 (2001-11-01), Galand et al.
patent: 6385199 (2002-05-01), Yoshimura et al.
patent: PCT/SE 00/02188 (2001-04-01), None
patent: WO 99/04522 (1999-01-01), None
patent: WO 00/49748 (2000-08-01), None
patent: WO 00/79762 (2000-12-01), None
Einarsson Torbjörn
Svanbro Krister
Jenkens & Gilchrist A Professional Corporation
Nguyen Linh V
Telefonaktiebolaget L M Ericsson (publ)
Tokar Michael
LandOfFree
Packet header compression using division remainders does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Packet header compression using division remainders, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Packet header compression using division remainders will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3015631