Layer two compression/decompression in a cellular communications network

0001The invention relates to the field of transmission of data frames in communications networks, and more particularly with transmission compression / decompression.

0002A number of communications networks are provided with compressors / decompressors loaded to compress the data in frames to be transmitted in order to increase their capabilities in terms of connections.

0003In these networks, the compression / decompression is called "layer one (1)" because it applies to the physical layer frames to compress all, regardless of their type. A compressive layer is to detect the overlap between the data contained in frames comprising identical portions of the stream of frames to be transmitted, said input frame, in order to remove any by signaling so that they can be reconstituted in decompression. The resulting compressed stream of frames are temporarily stored in memory buffer queues, for forwarding to destination based on the available transmission capacity.

0004We remind that the input frame is a frame that powers the compressor. It consists of a multiplicity of streams of frames constituting portions of initial frames associated with different transmission channels. For example, in TDMA transmission mode, each input frame is divided into a fixed number of time slots (or "time slots") each having one byte. This number is 32 in the case of E1 type frames, and 24 in the case of T1 type frames in accordance with G.703 / G.704 recommendations of the ITU-T. Usually, the frame frequency is 8 kHz, which allows an E1 frame to convey 31 information channels at 64 kbps (kilobits / s), at the rate of one channel per time interval.

0005Within a cellular network, the A-bis interface provides coupling between a base station controller (BSC for "Base Station Controller") and one or more base stations (BTS), using screens of Type E1 or T1. The A-bis interface supports two types of flows: the flow of traffic, speech or data types, and traffic flow. To carry traffic flows, each byte of the E1 or T1 frame is divided either into four doublets (two-bit sample called "nibble") each carrying a traffic channel to 16 kbps (mode called "full rate" or FR) or by eight half rate traffic channels (mode known as "half rate" or HR wherein each bit corresponds to a voice channel). The signaling flows are as generally carried by channels of two bits (channels 16 kbps) or 8 bits (64 kbps channels). An initial frame, E1 or T1 type and is a set of transmission channels, size of from 1 bit for channels 8 kbps to 8 bits for the 64 kbps channel; each channel for conveying the stream of frames, specific to each transmission channel.

0006Traffic flows are conveyed by TRAU frames (for "Transcoder / Rate Adapter Unit") exchanged between each BTS and the mobile switching center of the PLMN (for "Public Land Mobile Network"), in transiting the transcoder / unit rate matching (or TRAU). The latter is particularly intended to convert compressed speech data to 13 kbps digitized speech data to 64 kbps in order to make the speech channels compatible with the mobile switching center. The coupling between the mobile switching center and the TRAU is by an interface called A, while the coupling between the BSC and the TRAU is by an interface called A-ter.

0007In what follows, the term "active channel" all established traffic channel, ie for which a call setup procedure was successful via a signaling channel, procedure leading to the establishment of a traffic channel dedicated to that call, maintained until the end of call phase. A traffic channel is used to convey the TRAU frames exchanged between a caller and a called for the duration of the call, including during the silent phase.

0008It is further recalled that the TRAU frames may be of four types: speech frames (or data transfer) which comprise at least control data and useful data, silence frames which comprise at least data control, SID frames (for "Silence Descriptor") which comprise at least control data and useful data, and the bad frames (or "bad frames") which comprise at least control data.

0009In the following, the case of TRAU frames used for data transfer is a special case of TRAU frames used for the transport of speech, of the treatment being identical in both cases.

0010Patent Document <patcit id="pcit0001" dnum="EP1478195A"><text>EP-A-1478195</text></patcit> relates to a resource management device of a communication network with two levels of benefits.

0011When the decompressor receives the compressed stream of frames, it decompresses successively so as to reconstruct (or restore) the initial frames of which they constitute the portions. Such a mode of transmission with compression / decompression can be described as asynchronous. It introduces indeed transmission delays "end-to-end" for the initial frames (or flow) recomposed (e) s, which vary depending on the network load. In addition, these periods may vary significantly from one transmission channel to another. Moreover, the maximum transmission delay, which can be considered as a parameter guaranteed by the network operator, is high because the processing implemented throughout the transmission system is not deterministic.

0012This non-determinism untied end-to-end, combined with the fact that one must make comfortable margins for maximum transmission delay, make it almost impossible to use the compression technique presented above in a network voice communications, such as a GSM network.

0013The invention therefore aims to improve the situation, including ensuring a delay end-to-end on the transmission chain, while ensuring optimal compromise between compression time gain and said end-to-end, and if possible allowing compatibility of operation with a compression / decompression layer one.

0014It proposes to this effect an input frame data compression device consisting of the stream of frames defining portions of TRAU frames and signaling to be transmitted within a communications network and each comprising at least one header featuring representative monitoring data at least on their type and any useful data, including certain types of critical data and / or non-critical in terms of transmission delay end-to-end.

0015This compression device is characterized in that it is able to adapt its compression mode depending on the frame type. Such a device thus provides a so-called compression "layer two (2)".

0016More specifically, the device comprises, on the one hand, means adapted to analyze each frame header TRAU or signaling contained in input frames successively received, to determine its type, and secondly, means compression responsible for generating periodically (or cyclically) compressed frames to be transmitted, divided into first and second variable-sized sections, the first section having critical compressed data synchronously, and the second section having non-critical data compressed asynchronously.

0017In a mode of operation that can be called "fixed", the sizes of the first and second sections are variable and complementary.

0018The compression device according to the invention can comprise other characteristics that can be taken separately or in combination, including:<ul><li>the frame types may be selected from i) speech frames comprising in particular control data and useful critical data, ii) the silence frames comprising at least one of the critical control data, iii) SID frames comprising at least critical control data and non-critical user data, and iv) the bad frames including at least critical control data,</li><li>the compression means may be adapted to generate a compressed frame every N input frames received. They conduct an aggregation of the N input frames into a single frame compressed during a compression cycle,</li><li>the compression means may be responsible for integrating speech frames in compressed frames, without modification,</li><li>alternatively, when the speech frames include synchronization data, the compression means may be responsible for integrating speech frames in compressed frames, after deleting their data synchronization but without changing the control data and useful data,</li><li>the compression means may be, once they are in possession of all the monitoring data of a SID frame or silence, or a bad frame to remove these redundant information or insignificant to generate a word of relevant information, for example, two bytes,</li><li>the compression means may be compressed frames to form the first section comprises at least one word of relevant information and / or a speech frame without modification (or alternatively canceling synchronization data), and that a header identifying the start of the compressed frame and the information designating the input frame each channel received which owns the data from which the relevant information or data of the speech frame,</li><li>the compression means may be to extract useful data contained in the SID frames to store temporarily in a buffer, and signaling frames received, based on their order of arrival, to constitute a "linking" and then extracting the head of the sequence a portion whose size is selected according to the size variable in the second section to integrate it into the second section of compression of the current frame, accompanied by a header identifying the start of the extracted portion and information designating each of the input frame received channel which belong the user data or signaling data of that portion of sequence extracted, and finally to concatenate the second section to the first section to define a compressed frame to be transmitted,</li><li>the compression means may be adapted to adjust the respective sizes of the first and second sections depending on the network traffic in terms of speech frames and data transfer.</li></ul>

0019The invention also provides a data decompression device contained in compressed frames using a compression device of the type described above.

0020The decompression device is characterized in that it comprises, firstly, detection means responsible for analyzing the compressed frames to separate successively received their first and second sections, a second hand, first processing means adapted to reconstruct synchronously critical data initially contained in the input frames and having been compressed synchronously from the first received sections, thirdly, second processing means adapted to reconstruct asynchronously non-critical data initially contained in the input frame and having been compressed asynchronously from the second sections received, and fourthly, restitution means tasked with rebuilding the input frame from critical data and non-critical replenished.

0021Preferably, the means of synchronous processing and asynchronous processing means of such a pressure relief device operate in parallel.

0022The invention further provides a compressor / decompressor with a compression device and a decompression device of the type described above.

0023The invention finds a particularly interesting application, although not exclusive, in every point to point communications networks, point to multipoint and multipoint to multipoint, such as satellite communications networks, terrestrial communications networks, such as such networks on E1 or T1 base and the like, or based on Ethernet, IP, Frame Relay and ATM, and terrestrial communications networks using a satellite extension (or offset (s) satellite or terrestrial) over part their connections, such as cellular networks TDMA or CDMA.

0024Other features and advantages of the invention will appear on examining the detailed description below and the appended drawings, in which:<ul><li>the <figref idrefs="f0001">figure 1</figref> schematically illustrates a portion of a satellite extension communications network according to the invention,</li><li>the <figref idrefs="f0001">2</figref> schematically illustrates a frame processing system including an embodiment of a compressor / decompressor according to the invention,</li><li>the <figref idrefs="f0002">3</figref> schematically illustrates an example of E1 input frame,</li><li>the <figref idrefs="f0002">4</figref> schematically illustrates an example of initial TRAU frame,</li><li>the <figref idrefs="f0002">5</figref> schematically illustrates an example of a compressed frame of the invention, and</li><li>the <figref idrefs="f0002">6</figref> schematically illustrates an example of concatenation of non-critical user data stored in an insertion waiting buffered in a second section of a compressed frame of the type shown in <figref idrefs="f0002">5</figref>.</li></ul>

0025The attached drawings may not only serve to complete the invention, but also contribute to its definition, if appropriate.

0026The invention aims to allow the compression / decompression layer two (2) data frames in a communications network.

0027In what follows, we will consider, without limitation, that the communications network is a cellular network, such as a GSM (or 2G) or GSM / GPRS (2.5G), satellite extension (or offset (s) satellite). However, the invention is not limited to this type of network. As previously indicated, it relates to all types of communications networks point to point, point to multipoint or multipoint to multipoint, in which are defined the cable connections with fixed or shared transmission resources, or fixed transmission resources to radio links or shared, including satellite communications, terrestrial communications networks, such as networks E1 base or T1 and the like, or based ethernet, IP, Frame Relay and ATM, and terrestrial radio communication networks using a satellite extension (or offset (s) via satellite or terrestrial) over part of their connections, such as cellular networks TDMA or CDMA.

0028GSM communications network with satellite extension, of the type shown in <figref idrefs="f0001">figure 1</figref>May, on a very schematic but nevertheless sufficient for understanding the invention be summarized in a conventional radio subsystem, called a base station system (BSS "Base Station System"), coupled with a heart network or "Core network", here embodied, very reductive way, as a mobile switching center (MSC or "Mobile switching Centre") itself coupled to a PLMN (for "Public Land Mobile network "). The mobile switching center MSC is responsible for carrying out all the operations necessary to manage communications with user terminals UE.

0029Conventionally, the BSS comprises at least one CG traffic management center, for example arranged in the form of a "hub" H coupled to a transmitting / receiving station SER by which it will make satellite links Li with a relay satellite SAT communications. Alternatively, the hub H, which is the clearing with remote traffic stations ST can be separated and / or spaced from the management center CG.

0030SAT satellite relay is connected by satellite links to the traffic stations ST, the BSS, each having a base station called BTS (for "Base Transceiver Station") managing at least one radio cell in which user terminals EU may establish mobile communications. Each BTS base station is coupled to a processing system MT responsible for processing the frames to be transmitted and received frames and resource allocation, and itself coupled to a transmitter station / receiver SER 'through which takes place the satellite link Li with the satellite SAT.

0031We remind that the hub H is connected to different traffic stations ST by satellite links using a shared resource (Lf carrier) in forward mode (that is to say, the hub H to the traffic stations ST) and a set of carriers (Lr) shared by all links in Return mode (ie T traffic stations to the hub H).

0032The BSS system also comprises at least one base station controller BSC (or Base Station Controller) coupled, on the one hand, to the traffic management center CG, more precisely to its hub H, and secondly, the mobile switching center MSC. The BSC is primarily responsible for managing the resources of the various base stations BTS that are attached to it as well as the operating and maintenance functions of said base stations.

0033The sharing of transmission resources is controlled by the CG traffic management center. Specifically, the traffic management center CG dynamically allocates satellite resources in the form of transmission frequencies (operating mode SCPC) or time intervals (or "time slots") of a time frame (in mode TDMA operation (time division multiplexing)), according to the respective needs of different traffic stations ST it manages. In other words, the traffic management center CG allocates resources (or transmission channels) SAT satellite between different satellite links that it manages.

0034It is considered in the following, without limitation, the network is to TDMA.

0035As previously reported, the coupling between the BSC and base stations BTS is by the A-bis interface, synchronous type, which operates with a G.704 type screen (this is called E1 frames, the type shown on the <figref idrefs="f0002">3</figref>). As illustrated in<figref idrefs="f0001">figure 1</figref>The BSC can also be coupled directly, via terrestrial A-bis offsets, to independent BTS base stations of satellite links.

0036The coupling between the BSC and MSC mobile switching center is done here via a transcoder / rate adaptation unit (TRAU or for "Transcoder / Rate Adapter Unit"). The latter is designed to convert the speech data compressed to 13 kbps digitized speech data to 64 kbps in order to make the speech channels compatible with the MSC mobile switching center. It is recalled that in effect A interface between the MSC and the TRAU transcoder, data transmission and signaling in GSM / GPRS networks takes place in channels at 64 kbps.

0037As indicated above, the coupling between the mobile switching center MSC and the TRAU is made by an interface called A, while the coupling between the BSC and the TRAU is made by an interface called A-ter. In order to ensure a significant area coverage, the mobile switching center MSC can be coupled to several controllers BSC via several interfaces A-ter.

0038As is illustrated in <figref idrefs="f0001">2</figref>, The hub H, as each base station ST, MT includes a frame processing system comprising a modem 1 coupled to a compressor / decompressor 2.

0039The modem 1 comprises a modulator 3, producing a modulated carrier for modulating the frames compressed by the compressor / decompressor and 2 initially from the BSC or BTS, and a set of demodulators for each 4 demodulating the carrier from an ST or BSC remote traffic station to return the compressed frames to the compressor / decompressor 2 so that decompresses before transmitting them to the BSC or the BTS.

0040Each compressor / decompressor 2 has a compression device 5 and a pressure relief device (or expansion) 6. In practice, each frame processing system MT comprises as many compressors / decompressors that Li 2 satellite links. In fact, every compressor / decompressor 2 is supplied with demodulated and compressed frames by a demodulator 4, and feeds compressed frames modulating a modulator 3.

0041We consider in what follows that each compression device 5 receives input frames to compress, type E1 (but it could also be of T1 type frames, PDH frame type or frame carried by channels of SDH) from the BSC or BTS.

0042As is illustrated in <figref idrefs="f0002">3</figref>An E1 frame is divided into 32 time slots (or time slots) TS0 through TS31 each comprising one byte in accordance with G.703 / G.704 recommendations of the ITU-T.

0043Apart TS0 time slot, each time slot (TS1 to TS31) carries one or more transmission channels dedicated to communication (link) or transport of signaling. Each transmission channel carries a succession of "original frames" which are TRAU frames to transport speech and data transfer, or signaling frames from a BTS or BSC and to be transmitted the network or a user terminal UE.

0044It is recalled that TRAU frame can be of four different types:<ul><li>speech frames (or data transfer) which contain at least critical control data and useful critical data,</li><li>silence frames that contain at least critical control data,</li><li>SID frames which have at least critical control data and non-critical user data and</li><li>bad frames (or bad frames) that contain at least critical control data.</li></ul>

0045An example of a speech type of TRAU frame is illustrated in <figref idrefs="f0002">4</figref>. synchronization of the data consists of the first sixteen bits to zero (0), and nine of ten bits to one (1), an all of the following sixteen bits. Control data consists of bits C₁ -C₂₁ and T₁ to T₄. Specifically, the bits C₁ -C₁₅ constitute the header of the TRAU frame, bits C₁₆ -C₂₁ constitute a part of the end part of the TRAU frame, and the T bit₁ to T₄ (Optional) constituting another part of the terminal part of the TRAU frame, for the time shift (or "time alignment"). All control bits C₁ -C₂₁ characterizes the TRAU frame, including its type among the four types mentioned above. Furthermore, the field reserved to speech data (or "speech frame data field") is placed between the header and the end portion. speech data is useful data here.

0046A TRAU frame thus comprises 320 bits, or 20x16 bits, which correspond to a period of 20 ms (milliseconds) if "full rate" (FR) and "enhanced full rate" (EFR) or 160 bits is 20x8 bits in if "half rate" (HR).

0047In the case of a TRAU frame type silence the speech data field is invalid. In the case of a TRAU frame type SID, the speech data field carries the definition of comfort noise.

0048The case of the TRAU frame type is particularly bad frame. It corresponds to an initial frame received by a BTS with insufficient radio quality in terms of signal / noise ratio. One of the bits of control of such a frame indicates that poor quality (BFI bit, for "Bad Frame Indication").

0049In the example E1 frame illustrated in <figref idrefs="f0002">3</figref>The shaded parts are reserved for signaling frames, while the unshaded parts are reserved for the speech. In this example, given only for illustrative purposes, the time interval IT1 carries a signaling channel at 64 kbps, the time interval IT2 has a signaling channel to 16 kbps and 3 traffic channels at 16 kbps (or 6 channels traffic to 8 kbps, or even a combination of channels to 16 kbps and 8 kbps channels), and the interval time IT3 only carries traffic channels.

0050A compression device 5 ensures not only the compression of frames to be transmitted, but also the adaptation of data unit sizes at the interface with the modem 1, which is of type E1, T1, Ethernet, IP or ATM. Furthermore, a decompression device 6 ensures not only the restitution, by decompression, the initial frames (that is to say as they were before being compressed in the traffic station ST), but also the adaptation at the interface with the modem 1, which is generally the same type as that of the compression portion.

0051As noted above, certain flows, such as those containing speech data, are critical in terms of transmission delay end-to-end, and therefore must be treated differently from other streams in order to allow the operator to ensure their transmission by a time limit fixed end-to-end and as short as possible.

0052The invention therefore proposes to address critical flow in a deterministic synchronous processing throughout the chain of transmission (compression - transmission - decompression), and flows not criticism asynchronous processing (non deterministic) all along the chain transmission.

0053The dissociation between streams processed synchronously and asynchronously is established by configuration. Specifically, it is the operator that specifies which channels are to be processed synchronously (usually those that support voice communications), and which channels must be processed asynchronously (usually those that support the signaling information).

0054The compression device 5 is configured to differentiate the critical flow of non-critical flow, then they apply different treatments (synchronous and asynchronous).

0055It includes for this purpose an analysis module 7 to analyze the control bits of each TRAU frame received from the BTS or BSC to determine the type of the received frame. This analysis module 7 operates by compression cycle. In the example described, the compression cycle involves N = 64 E1 frames, a time of 8 ms, each E1 frame having a duration of 125 microseconds (microseconds). The compression cycle defines the rate at which the compressed frames TC is transmitted from the compression device 5 to the decompression device 6 at the other end of the link. The compression cycle also sets the rate at which input TRAU frames are analyzed.

0056An initial TRAU frame having a duration of 20 ms, it takes 3 consecutive compression cycles to have all the information it conveys, including control bits located at the beginning and end of TRAU frame. Depending on the value of the control bits, the TRAU frame is identified as speech frame (or data transfer frame), silence frame, SID frame or bad frame. The analysis module 7 can determine the type of compression to be applied to the TRAU frame received during the 3 compression cycles considered.

0057For the signaling channels, the analysis module 7 merely extract the frames carried by the channel, and to store them in a buffer (or buffer) dedicated to the channel concerned.

0058The compression device 5 also includes a compression module 8 responsible for generating at each pressing cycle a compressed frame TC, using data extracted by the analysis module 7.

0059Specifically, when the compression module 8 is informed by the analysis module 7 of the type of TRAU frame, it transmits significant data carried by the TRAU frame:<ul><li>payload control and data critical to a recognized frame as a speech frame or data transfer,</li><li>control data for a silence frame,</li><li>control data and non-critical data useful to the recognized frame as a SID frame,</li><li>control data for a bad frame.</li></ul>

0060For the signaling channels, the compression module 8 multiplexes signaling flow with traffic flows.

0061According to the invention, and as illustrated in <figref idrefs="f0002">5</figref>Each TC frame, compressed by the compression unit 8 comprises first S1 and second S2 sections of varying sizes. The first section S1 includes critical data compressed synchronously, and the second section S2 comprises non-critical data compressed asynchronously.

0062Preferably, critical data is data (or bits) of control (C₁ -C₂₁ and T₁ to T₄) And the speech data, while the non-critical data are the comfort noise data (SID frames) and signaling data (signaling frames).

0063Compression can be carried out as indicated below.

0064The compressor 8 is first of all the first sections S1 compressed frames TC, also called synchronous frames. These synchronous frames S1 include, as shown in<figref idrefs="f0002">5</figref>A header EN1 and a ZIP field and / or a ZD1 field.

0065The ZIP field is a field dedicated to the said relevant information contained in the control data C₁ -C₂₁ and T₁ to T₄ (25 bits) of the SID frames, silence frames and bad frames. It preferably comprises a word of relevant information (or RI 1-16 for "Relevant Information"), for example, two bytes (16 bits), formed by removing redundant information or insignificant contained in the 25 bits of data control.

0066The ZD1 field is a field dedicated to speech data in the speech frames. It includes, for example, all data contained in the speech frame, ie the synchronization data, control data and speech data. But, alternatively, it may contain a compressed version of the speech frame, that is to say its only control data and speech data (synchronization bits (35 bits) are then deleted).

0067The header field EN1 identifies the beginning of the synchronous frame and thus the beginning of the compressed frame. It also includes information designating the input frame each channel received which owns the data from which the relevant information of the two-byte word (ZIP) or the speech frame data (ZD1). In fact, it includes all the necessary information to the decompression device 6 to restore critical data and therefore the frames from which they arise.

0068Preferably, and as shown, the ZIP field is inserted between the header EN1 and ZD1 field.

0069Once the first section consisting S1, the compressor 8 is the second section S2 compressed frames TC, also called asynchronous frame. As illustrated in<figref idrefs="f0002">5</figref>Each asynchronous frame S2 EN2 has a header and a field ZD2.

0070The ZD2 field has non-critical data, specifically the signaling data and / or user data SID frames of the comfort noise. It may also be at least partly made of padding data, in the absence of non-critical data to be transmitted or when the number of non-critical data to be transmitted is insufficient.

0071In fact, these non-critical data is temporarily stored by the compression module 8, as and when they arrive in buffers queues where they are sequences of the type shown in <figref idrefs="f0002">6</figref>.

0072This data is then extracted by the compression module 8 according to their order of arrival (oldest, that is to say, those placed at the head of the chain, are extracted first), but mostly based size available at the time of compression in the second section S2.

0073The respective sizes of the first S1 and second S2 sections of a compressed frame TC preferentially vary independently, with a fixed or progressive overall value, depending on the network load (in terms of number of voice frames to be transmitted).

0074With fixed transmission resources, the number of voice frames to be transmitted is high, such as during peak hours, the larger of the first section S1 and the greater the size of the second section S2 is small. Conversely, less the number of voice frames to be transmitted is high, such as off-peak hours, plus the size of the second section S2, the greater the size of the first section S1 is small.

0075With scalable transmission resources, which is the case with a transmission device operating DAMA mode, the two sections can move independently one vis-a-vis each other, the transmission device adapting the resources allocated to the global burden constituted by the stream conveyed through the first section S1 and S2 via the second section.

0076The respective sizes of the first S1 and second S2 sections are updated at each compression cycle in particular according to the traffic load. Thus, each time the compressor 8 knows the amount of non-critical data that it can extract a sequence stored in a buffer for supplying the field ZD2 of the asynchronous frame S2.

0077When the number of data stored at the head of a sequence and belonging to the same frame, for example signaling, is greater than that available (in progress) in the second section S2, only a portion of these data is integrated into the second section S2, the other party pending integration in the next compressed frame. It is also conceivable that the number of data stored at the head of a chain belonging to the same frame, is lower than that available (in progress) in the second section S2. In this case, these data are integrated into the second section S2 concatenated with some (or all) of the following data in the sequence and belong to another frame.

0078The header field EN2 identifies the beginning of the asynchronous frame S2. It contains information that designate each input frame of the channel received which owns the data from which the data in the field ZD2, and the length of said field ZD2. In fact, the EN2 header field includes all necessary information to the decompression device 6 to reconstitute the non-critical data and therefore the frames from which they arise.

0079Once the second section S2 is formed, the compression module 8 concatenates the first section S1, which is a compressed frame TC, of ​​the type illustrated in <figref idrefs="f0002">5</figref>. The compression device 5 can then put the data block format at the interface with the modem 1, before passing to the modulator 3 concerned.

0080The delay introduced by the asynchronous compression varies depending on the capacity offered by the transmission link at a given time and capacity used by synchronously treated flows, while the delay introduced by synchronous compression is fixed.

0081The operation of the decompression device 6, according to the invention, is roughly reversed from that of the compression device 5 which has just been described.

0082Specifically, the decompression device 6 firstly includes a detection module 9 to analyze compressed frames TC, it receives (in succession) the demodulator 4 which it is coupled to separate their first S1 and second S2 sections. It detects to what their respective EN1 and EN2 headers.

0083The decompression device 6 also comprises a first processing module 10 to reconstruct loaded synchronously from the first sections S1 received, critical data which were initially contained in the input frames and which has been compressed synchronously.

0084Specifically, the speech frames are reconstructed using the useful data in the ZD1 field (after a possible reintegration synchronization data). They are then transmitted to a frame restoration module 11, the decompression device 6, in charge of restoring them to their original location in the E1 input frame, given the information in the header EN1.

0085The silence frames are reconstructed using the words of relevant information contained in the ZIP field and restored by the frame restoration module 11 to their original location in the E1 input frame, given the information contained in 'header EN1 and the default value of the speech data field.

0086The reconstruction of SID frames is a little more complex, since it requires a reconstruction performed by a second processing module 12 of the decompression device 6, dedicated to the compressed stream transmitted asynchronously. The processing module 12 stores the content of frames received via the asynchronous transmission channel (second section S2 of the compressed frame TC) content relating to SID frames or signaling frames. For content related to SID frames, the latest content replaces the previous contents; this content is the useful field D1-D260 of the SID frame, characterizing the comfort noise for the given channel. The second processing module 12 has at least as buffers (or buffers) as traffic channels. To channel D1-D260 received data to the desired traffic channel, the second processing module 12 operates the EN2 header information to determine which channel belongs D1-D260 field received, and therefore what buffer (or buffer) data should be stored.

0087In parallel with this process of storing data received asynchronously, the first processing module 10 reconstructs the body of the SID frame from the words of relevant information, received synchronously, by following the same process as for the frames silence. The body of the SID frame relates to the bit synchronization portion, check bits C₁ -C₂₁And time alignment bits T₁ to T₄.

0088As for speech frames and silence frames, the frames of the playback module 11 is responsible for each SID frame return to its original location in the frame output E1. The SID frame restored by the frame restoration module 11 is formed by the body of the SID frame, reconstructed by the first processing module 10, and by the useful field D1-D260 of the SID frame, buffered by the second module 12 treatment.

0089For the first frame of silence which follows a succession of speech frames, the second processing module 12 can sample data characterizing the comfort noise in the latest received frame (SID frame) transmitted during the transition from speech mode to silent mode, using the same method that the speech frames. This ensures a more accurate transition of the synchronous transmission mode, which applies to speech frames in asynchronous transmission mode, which applies to SID frames: comfort noise present in the buffer (or buffer) of channel is considered newer than the one that would have been present without this transaction, namely the sampled sound just before the sequence of speech frames. With this operation, the noise present in the buffer (or buffer) is just noise sampled at the end of the speech sequence, even before an asynchronous transmission of data D1-D260 has been made for the same channel.

0090It is important to note that the second processing module 12 is also responsible for reconstituting the signaling data contained in the field ZD2. They are then transmitted to the frame restoration module 11 so that it restore them to their original location in the E1 input frame, given the information in the header EN2.

0091Preferably, first 10 and second 12 processing modules operate in parallel.

0092Under offsets by radio or satellite, the invention also aims to ensure that the error rate, which is guaranteed for every type of stream meets the constraints of the flow concerned.

0093As known to those skilled in the art, some modems are arranged to carry out an encoding of the information contained in the frames to be transmitted. This encoding is fixed and independent of the types of transmitted streams. It limits the error rate in reception by inserting redundancy data on the transmitted information. To ensure optimum spectral efficiency, that is to say maximizing throughput in fixed satellite band, the rate of this encoding is set at a low value, so that the overhead introduced by the redundancy data remains the most low as possible, for a given error rate. The tolerable error rate on an A-bis offset with compression depends on the conveyed stream, for example:<ul><li>error rate generally less than 10^-10 for the header information necessary for the decompression process,</li><li>low error rate also for speech frames of the order of 10^-8,</li><li>low error rate for signaling frames,</li><li>average error rate for the data relating to comfort noise.</li></ul>

0094To optimize the transmission based on the transmitted flux, the invention provides a pre-encoding for making the error rate for every type of stream (or traffic).

0095To this end, the MT processing system includes an on-encoding module 13, preferably located in the compressor / decompressor 2, and responsible for pre-encode independently each stream type, except preferably the most tolerant flux in error rates, such flows relating to comfort noise.

0096This pre-encoding is to introduce redundancy bits transmitted for each type of flow, except for flows requiring quality lowest transmission (comfort noise in the previous example). The number of redundancy bits depends on the desired quality. redundancy bits are used in reception to detect and correct errors introduced on the link. The more redundancy bits, the higher the error rate can be reduced with this principle, given link quality (signal to noise ratio and intermodulation).

0097pre-encoding the term refers to the introduction of additional redundancy bits, vis-à-vis those introduced by the transmission device, which makes it also an encoding, the same principle, to reduce the impact of imperfections in the bond, however following an undifferentiated use between the different types of directed flows. The flow separation performed at the compressor allows pre-encode each stream based on the criticality of the data conveyed.

0098For this purpose, the on-encoding module 13 is supplied with compressed frames TC by the compression device 5 and supplies the compressed frames and the pre-encoded modulator 3 at which the compressor / decompressor 2 is coupled. The modulator 3 then has only to perform classical overall encoding compressed frames and pre-encoded.

0099In reception, the demodulator 4 performs the correction concerned classically overall error on the compressed frame TC received, he has previously demodulated, then the on-encoding module 13 applies to the compressed frame and a clean demodulated désencodage each type flow.

0100The overall error rate is well suited to each type of flow for data transmission conditions, that is to say for a signal to noise ratio E_b/NOT₀ and given a given intermodulation rate. This will significantly increase the spectral efficiency of MT treatment system, reducing the redundancy rate applied to the overall flow, the most critical flow being provided for their own protection.

0101The compression device 5, and in particular analysis modules 7 and 8 of compression, the decompression device 6, and in particular its sensor modules 9 and restitution 11 together with the first 10 and second 12 processing modules, and the module over-encoding 13 may be constructed as electronic circuits, software modules (or computer), or a combination of circuits and software.

0102The invention combines the advantages of synchronous processing and those of asynchronous processing. It also enables a deterministic delay end-to-end and invariant for channels that the operator regards as critical, especially for the transmission of speech.

0103In addition, the invention allows a delay end-to-end short-lived, and configurable to compromise with the compression overhead.

0104In addition, the invention provides a smoothing of asynchronous streams because they are transmitted using the remnants of bandwidth (second sections of the compressed frames).

0105Also thanks to the invention, each traffic type can be considered.

0106Finally, the invention is compatible with a method of allocating resources (or transmission channels) of the satellite between different satellite links, called DAMA (for "Demand Assignment Multiple Access"), through a comprehensive flow slowly varying (as opposed to burst traffic (or "bursty")).

0107On this last aspect, the invention takes advantage of the dynamic characteristics of the transmitted signals: typically 80% of traffic consists of voice traffic. Out, word of the traffic moves slowly, when a new call is established, or when a call ends or when the floor gives way to silence on a given channel, and vice versa. These events occur at a slower pace compared to applications multimedia transmission systems where data flows are sporadic, especially whenever a user requests a file transfer or accesses a new web page. In addition, for a high number of concurrent calls, which is the case during peak hours, there is a global phenomenon smoothing of all speech traffic, around the median of traffic between 35% and 50% of the established channels in accordance with the time average proportion of speech vis-a-vis of the communication time. At the busiest time, and therefore more burdensome for the transmission device, the charge related variations are reduced through this smoothing phenomenon, suitable therefore compression means of operation associated with a transmission system operating DAMA Mode: system responsiveness can be relatively slow, traffic moving slowly.

0108The invention is not limited to the compression device embodiments, pressure relief devices, compressor / decompressor and communications network described above, only as examples, but encompasses all the variants that may consider the skilled in the art within the scope of the following claims.

0109Thus, the invention also relates to other than cellular communication networks. The principle of synchronous / asynchronous mixed transmission with self-adaptation of the transmission resources used for each flow makes it possible to optimize the type of traffic offsets speech, data and / or signaling on other than cellular networks, based on the relative criticality of the conveyed stream, and in particular ensuring low and deterministic time for the most critical flow.