WO2004109986A1

WO2004109986A1 - Optimised communication method, access device and system for carrying out the same

Info

Publication number: WO2004109986A1
Application number: PCT/FR2004/001321
Authority: WO
Inventors: Thomas Noel; Jean Lorchat
Original assignee: Universite Louis Pasteur; Centre National De La Recherche Scientifique
Priority date: 2003-05-28
Filing date: 2004-05-27
Publication date: 2004-12-16
Also published as: FR2855698A1; FR2855698B1

Abstract

The invention relates to a method for communication by transmission of frames or packets of data, using at least one first and one second communication network, connected to each other by at least one interface or access point, the at least two networks accepting different maximum lengths or sizes of frames or packets. The method is characterised in generating at the point of access or each point of access (PA), an aggregation of frames (T1) for the messages transferring from the first network (PR) with a lower maximum frame size to the second network (SR) with a higher maximum frame size and a disaggregation of frames (T2) for the messages transferring from the second network (SR) with a higher maximum frame size (T2) to the first network (PR) with a lower maximum frame size, followed by a reconstitution of the frames (TI) having previously undergone an aggregation operation.

Description

Optimized communication method, access device and system for its implementation

The present invention relates to the field of communication, more particularly the communication being carried out on the basis of data transmissions by frames or packets through at least two networks linked together and having different maximum frame sizes.

The subject of the present invention is, in the abovementioned context, a communication method optimized in terms of size of transmitted frames, an access device for implementing this method and a system implementing this method. Currently, during the transmission of a data stream between a first network and a second network, accepting different maximum frame sizes, the frames are formatted so as to be able to circulate on the network having the most authorized maximum frame size most weak, these frames passing to the other network and circulating on the latter without modification or alteration.

This results in a low filling rate of the frames circulating on the network having the largest authorized maximum frame size and therefore of the reports [quantity of data transmitted / congestion of the network or solicitation of the equipment of the network] and [quantity of data transmitted / energy consumption of the equipment used] very unfavorable, and far from optimal.

The harmful influence of the situation described above is particularly penalizing for wireless transmission equipment, in particular those with autonomous power supply, due to a consumption of energy that is not proportional to the quantities of data transmitted.

The present invention aims, if not to completely overcome the aforementioned drawbacks and limitations, at least to significantly limit the harmful influences.

To this end, the main object of the present invention is a method of communication by transmission of frames or data packets, using at least a first and a second communication network linked together by at least one interface point or access, said at least two networks accepting lengths or sizes maximum of different frames or packets, method characterized in that it consists in carrying out, at the level of said access point or of each access point, on the one hand, an aggregation of frames for the messages passing through the first network with a lower maximum frame size to the second network with a higher maximum frame size and, on the other hand, a disaggregation of frames for messages passing from the second network with a higher maximum frame size to the first network with a lower maximum frame size , followed by a reconstruction of the frames which were the subject of an aggregation operation beforehand. The saving in terms of energy and congestion is achieved by reducing the volume of information passing through the network, since the overload involved in sending several frames is reduced to the overload necessary for sending a frame. In addition, in the specific case of radiofrequency transmission, the cost per bit decreases when the size of the frame increases, for reasons related to the nature of radio communications, which makes it possible to realize additional savings when several frames are grouped together. into a larger frame.

The aggregation or grouping of smaller wefts into larger straw wefts can, in the simplest embodiment of the invention, be carried out by simple juxtaposition, without modification of any of said smaller frames, and respecting the maximum size allowed for larger frames.

However, according to another embodiment, making it possible to achieve a more efficient grouping of frames, provision may be made for the frame aggregation operation to be preceded by an operation for removing redundant data between the different frames of each group of frames circulating on the first network and intended to form a frame capable of circulating on the second network, and the operation of reconstituting, in their entirety or not, frames intended to circulate on the first network following a disaggregation of frames circulating on the second network, is preceded by an operation to regenerate redundant data deleted during the aggregation.

Preferably, the data deleted during the aggregation and regenerated during the reconstruction consist of identical data at the level of the headers of the frames to be aggregated.

In the context of this second variant, it is also advantageously provided that the frames circulating on the second network and formed by the aggregation of several frames compatible with the first network contain additional information concerning the structure of the corresponding aggregations.

In view of the above, at least three alternative embodiments of the invention can be envisaged in relation to the organizations and the equipment of current communication networks.

Thus, according to a first variant, the method can consist in aggregating or grouping frames of the same type or nature and whose destinations in the second network are identical, this without modification or alteration of said aggregated frames.

According to a second variant, the method may consist in aggregating or grouping frames of the same type, originating from the same source and whose destinations in the second network are identical, at least part of the header of each frame to be transmitted on the second network being common to all the frames aggregated or encapsulated in this frame.

Finally, according to a third variant, the method can consist in aggregating or grouping frames of the same type, forming part of the same message or of the same data stream, originating from the same source and whose destinations in the second network are identical. , at least part of the header of each frame to be transmitted on the second network being common to all the frames aggregated or encapsulated in this frame.

Each of the above variants groups the frames differently, in an increasingly specific way, but also more and more efficient in terms of gains, by progressively reducing the redundancy contained in the packets which have been grouped.

The method will find particular application when the first network is a wired network, in particular local, and that the second network is a wireless network or with radiofrequency transmission, but also when the two networks are wireless networks or with radiofrequency transmission.

A completely preferential application of the method according to the invention can be envisaged when the transmission in the second network meets the communication standard known under the designation IEEE 802.11. In connection with this last standard and with the three variant embodiments mentioned above, it is possible to provide three concrete mechanisms for grouping or assembling frames. Thus, a first mechanism (hereinafter referred to as level 2 or two) groups (before transmitting them) the frames arriving on the access point bound for the wireless equipment without any modification, and encapsulates them bound for the wireless equipment. This then splits the incoming frame to reconstruct the original frames.

A second mechanism, hereinafter referred to as level 2+ or two +, groups (before transmitting them) all the frames for which the header of MAC level is common or identical (i.e. frames of the same type and with destination of the station, from the same level 2 source). This means that for all the frames which will be assembled in only one, a single header will be kept, which saves space in the final frame. This particular case applies very well in the case of connections to a large-scale network such as the Internet since all traffic coming from outside the wireless network will have as MAC level address the address of the router interface. connected to the wireless network. The reconstruction of the frames is done by taking the common header and concatenating the differentiated parts of the frames, which are assembled behind the common header at MAC level.

Finally, a third mechanism (hereinafter referred to as level 3 or three) extends the scope of the headers common or identical to the information of the network level protocol (for example IP (Internet protocol), in its IPv4 or IPv6 version), this which saves a little more space in the final frame. However, for the frames to have a common header at the IP level, these must belong to the same data flow (that is to say that the applications which exploit the data contained in the frames the same). This scenario is intended for high speed multimedia streams used to transport high quality audiovisual information.

The three mechanisms described above apply to communications from the access point to wireless equipment, but they are completely reversible and therefore allow energy savings in both directions of communication.

The present invention also relates to an interface or access device intended to ensure a link between a first and a second communication networks authorizing the circulation of frames or data packets, said two networks accepting maximum lengths or sizes different frames or packets. This device is characterized in that said device comprises means capable of carrying out, on the one hand, an aggregation of frames for messages passing from the first network with a lower maximum frame size to the second network with a higher maximum frame size, and on the other hand , a disaggregation of frames for messages passing from the second network with a maximum maximum frame size to the first network with a maximum maximum frame size, followed by a reconstruction of the previously aggregated frames.

Preferably, this device will be suitable for implementing the communication method described previously. Said device can be integrated into a mobile or cellular telephone and / or into an access point unit ensuring the connection between one or more stations of a wireless communication network and a wired network.

Finally, the invention also relates to a communication system comprising at least three separate communication networks all authorizing the circulation of frames or data packets, one of said networks, called the central network, accepting frames of length or size. maximum greater than those accepted by the other two networks, and being able to connect the latter two networks by means of communication channels or paths established between at least two interface or access points ensuring respectively a connection between said central network and one of the other two networks, this system being characterized in that it implements the communication method described above at each interface or access point.

The invention will be better understood from the following description, which relates to a preferred embodiment, given by way of nonlimiting example, and explained with reference to the appended schematic drawings, in which: FIGS. 1A and 1B illustrate the frame structures of the type

Ethernet and the type corresponding to the IEEE 802.11 standard;

Figures 2 to 4 respectively illustrate the frame aggregation mechanisms called in this level 2, level 2+ and level 3; FIG. 5 illustrates, put in parallel, the aggregation and disaggregation operations of the three aforementioned mechanisms; FIG. 6 represents the discharge curve of the second battery of a mobile during the transmission of the latter to the fixed network (with PSP) with and without implementation of the method according to the invention;

FIG. 7 represents the discharge curve of the second battery of a mobile during the reception of data (without PSP) with and without application of the method according to the invention;

FIG. 8 represents the discharge curve of the main battery of a mobile during transmission to a fixed network with and without application of the method according to the invention, and, FIG. 9 is a schematic representation of a communication system according to the invention.

As mentioned previously and as is apparent from FIGS. 2 to 5 and 9 of the appended drawings, the communication method concerned uses at least a first PR and a second SR communication networks linked together by at least one interface point or access PA, said at least two networks PR and SR accepting different maximum lengths or sizes of frames or packets.

According to the invention, said method essentially consists in carrying out, at the level of said access point or of each access point PA, on the one hand, an aggregation of frames Tl for the messages passing through the first network PR with size of lower maximum frames to the second SR network with higher maximum frame size and, on the other hand, a disaggregation of T2 frames for messages passing from the second SR network with higher maximum T2 frame size to the first PR network with frame size lower maximum, followed by a reconstitution of the Tl frames which have previously been the subject of an aggregation operation.

For the implementation of this method, the invention provides in particular at least one interface or access device intended to ensure a link between a first PR and a second SR communication networks authorizing the circulation of frames or packets of data, said two networks accepting different maximum lengths or sizes of frames or packets. This device comprises in particular means capable of carrying out, on the one hand, an aggregation of frames T1 for the messages transiting from the first network PR with a lower maximum frame size to the second network SR with a higher maximum frame size, and on the other hand On the other hand, a disaggregation of T2 frames for messages passing from the second SR network with a higher maximum frame size to the first PR network with a lower maximum frame size, followed by a reconstruction of the previously aggregated T1 frames.

The invention will now be described in more detail on the basis of a particular, non-limiting embodiment, based on infrastructure networks meeting the IEEE 802.11 standard and in connection with mobile communication devices (such as for example cell phones or similar), associated with wireless networks comprising in particular a plurality of stations distributed geographically over a dominated territory and wired networks of the Ethernet type. As mobile devices spread more and more and are used to communicate, most of them can now be equipped with Wireless Network Interface Cards (NICs). Advanced technology in packet radio networks operating with the industrial, scientific and medical band is known as IEEE 802.11. Given the high power consumption of NIC cards that follow this standard, the media access control layer (MAC) of the IEEE 802.11 specification has been designed to allow stations on standby to be mostly awake or asleep when used in combination with an access point (AP). This operating mode is called "infrastructure mode" and serves as a basis for the mechanisms of the invention.

PA points send periodic messages known as beacons, allowing stations to locate them and move freely from one PA point to another. In addition, the beacon message is also used to interrogate dormant stations and deliver incoming messages to them which have been buffered between the beacon messages. This operating mode of the NIC card is known as interrogation with energy saving (PSP) and makes it possible to obtain great energy savings since the station can be dormant between the beacon messages. The main drawbacks of this mode are the high latency introduced by the buffering process (around 300 milliseconds) and the lower maximum throughput to the dormant station.

FRAME AGGREGATION The inventors noticed the maximum frame size first of all that the maximum transfer unit (MTU) defined in the IEEE 802.11 specification (2312 bytes) is greater than that of the system Ethernet (1,500 bytes), which is used to transport data from the fixed network to the PA point. So there is basically more room for data in an IEEE 802.11 frame, this extra room is only being used due to the fragmentation of IP protocol packets coming from the wired network over the Ethernet link, as is apparent when comparing the Figures 1A and 1B. In addition, we can easily see (see in particular C. Fraleigh, S. Moon, C. Diot, B. Lyles, and F. Tobagit, "Packet level traffic measurements from a level 1 IP backbone", Sprint ATL, Tech. Rep., 2001) that most IP packets that are transported on the backbone use sizes less than 800 bytes. This ensures that aggregation will be possible and efficient. In addition, the operating system can be tricked into building packets of arbitrary size by setting the MTU most suitable for the wireless interface.

From an operating system perspective, IEEE 802.11 NICs are considered Ethernet NICs, and the NIC card driver encapsulates Ethernet frames over the wireless link. The key mechanism for frame aggregation is to add consecutive frames together and send them as a single Ethernet frame, reducing the second level network service load introduced by IEEE 802.11 encapsulation. A single IEEE 802.11 header is used for all aggregated frames. The service load of the first level introduced by the physical layer itself (such as the preamble and the frame delimiter) is also reduced in passing, since a single preamble is necessary for the whole aggregated frame. On the receiver side, this frame is separated into the various Ethernet frames, which are retrieved via the NIC card driver at the upper levels of the network stack.

Using the above method, any frame can be aggregated with any other frame, because in infrastructure mode, the source address and the destination address IEEE 802.11 are always identical for a given destination station because of that the point PA delivers frames to the station. But by aggregating frames by their second level addresses we can further reduce the network service overhead by keeping only one of the Ethernet headers, and using it as a common factor. Finally, frames can be selected by their third header information level by allowing another further reduction of the network service load.

On the basis of the observations above, it is possible to define three types corresponding to three levels of aggregation with different conditions on the headers of the incoming frames.

The first level called "level two" can incorporate any pending frame into a single frame, but simply by adding it entirely to the main payload (see Figure 2). Since the entire frame is preserved, including the second level header, there is no special processing required and the frame is encapsulated, without any modification of its content.

The second level called "level two +" can only incorporate frames with corresponding second level headers (i.e. the source address and the Ethernet destination address are the same, which is actually likely in large networks using routers). The resulting single frame (see Figure 3) contains the common header, which is followed by all the payloads of the grouped frames joined together in a single payload. Because the headers are separated, there is a need for some additional frame processing. The second levels must be matched with each other to obtain a common sequence. These headers are separated and only one is kept so as to reconstruct the frame on the destination station.

The last level, called "level three" requires the most processing because it maps the frames to be grouped by the content of their third level header. Therefore, the mutually aggregated frames belong to the same flow: the source and destination addresses of the IP protocol correspond, as do the remaining fields of the IP header. This level of aggregation (see Figure 4) imposes filtering on the third level header and is the most complex among the three principles of aggregation.

The reconstruction of the frames takes place by looking at the header of the main payload, consisting of two identifiers: the level of aggregation used by the main payload, and the other, the number of aggregated frames included. in this payload. Each payload frame begins with an identifier representing its size, followed by the specific part of the frame itself. Using level two aggregation there is no processing required and the frames are extracted as is. For level two +, the frames are reconstructed using the Ethernet header of the first frame and the specific payload of the aggregated frames. Finally, the aggregated frames of level three are reconstructed using the Ethernet header and the IP header of the first frame, the following frames being extracted in the main payload, only the third level payload remaining.

The inventors have carried out a theoretical analysis of the benefits expected or obtained by the aggregation mechanisms described above, with regard to the network load and the energy consumption. In particular, they analyzed the theoretical gains induced by the combination of the additional Ethernet frames (that is to say all the aggregated frames except the first) compared to their transmission in the form of isolated radio frames.

A. Network service load savings With regard to network load, the obvious advantage of aggregation is the reduction in service load due to the smaller amount of radio frames transmitted over the medium. Each aggregated frame saves an IEEE 802.11 header, as does the physical preamble required for the radio frame to be transmitted over the wireless channel.

When using the principle of the two + level mechanism, each additional frame which is encapsulated, saves an Ethernet header. Finally, in the principle of the level three mechanism, a header of the IP protocol is saved at the same time as the Ethernet header for each additional frame.

When using this last principle, the reduction due to the header depends on the version of the IP protocol used [ie IPv4 [J. Postel, "RFC 791 - Internet protocol specification", 1981], or IPv6 [S. Deering and R. Hinden, "RFC 2460 - Internet protocol version 6 specification", December 1998.]). The main difference between the two versions is the address space which is four bytes for IPv4 and sixteen bytes for IPv6. Using IPv4, 20 bytes are saved by the latter principle, while 40 bytes can be saved with the IPv6 protocol. The network load resulting from a flow with different aggregation levels is obtained by evaluating the following formula, according to these conditions: If represents the packet size of the IP protocol (including the head); ^ the size of the IP header; Ej the size of the Ethernet header, and W _x the size of the IEEE 802.11 header. The parameter t represents the number of packets to be sent, and the parameter N ₂ , N _2p , N ₃ the number of packets incorporated in each radio frame using the aggregation levels two, two + and three, respectively.

Using no aggregation process, each packet goes through each level of the network stack, and is concatenated with an Ethernet header and an IEEE 802.11 header. This constitutes the worst case when an aggregation is not validated, and it is used as a basis to be able to assess the point at which the network load is saved by the method according to the invention. The total load generated at the MAC command level (preamble to the physical layer not included) is given by:

∑ (W, + E _I + S _I ) ι = 0

As soon as the aggregation process is validated, the size of the radio frame increases, and several packets can be made up into a single radio frame, depending on the size of the packet. By using level two aggregation, the load can be reduced by saving spurious IEEE 802.11 headers. The new formula for calculating the total load becomes

ΣCW. + N ^ CE. + S,)) ι = 0

TABLE I NETWORK LOAD SAVINGS

By using a level two + aggregation principle, additional Ethernet headers are saved by using the common header of all Ethernet frames. The total network load decreases again, as can be seen in the new formula: ∑ (W, + E _I + N _2p * S _I )

1 = 0

The last level of aggregation provides different savings in network load depending on the version of the IP protocol used, because the address space differs between the IPv4 and IPv6 protocols. The size of the header appears in the form Hj in this last formula indicating the total network load generated by a flow formation using level three aggregation.

'∑ (W ₁ + E _I + H ₁ + N ₃ * (S _I -H ₁ )) i = 0

The following table I summarizes the reduction in network load for additional packets, for all possible principles. It presents the gross load savings obtained by using a common header for each additional frame, so that the final savings differ slightly, since the aggregation adds some service load used by the destination station to be able to reconstruct the frames.

B. Example

In order to obtain numerical results, the inventors have assumed the formation of streams of fixed size using third level payloads of size 512 bytes. This stream is sent at 512 kilobytes per second (which is roughly equivalent to a high quality video stream), for 15 minutes. The network load resulting from this flow is obtained using the formula defined in paragraph IV -A.

C. Energy savings

Given the manner in which radio communications are obtained using the IEEE 802.11 standard, the inventors expect that the cost per bit transmitted will decrease as the length of the radio frame increases. This was notably exposed in JP Ebert, B. Burns, and A. olisz, "A tracking-based approach for determining the power consumption of a wlan network interface,", in Proc. Of European Wireless 2002, Florence, Italy, February 2002, pages 230 to 236 and forms the basis of our energy saving analysis since the aim of aggregation is to increase the length of radio frames by combining together multiple consecutive Ethernet frames.

But by combining these Ethernet frames into a single radio frame, redundant information is factorized and overall network traffic is reduced. Therefore, fewer bytes are sent using larger radio frames and exponential savings can be expected using higher aggregation principles. This means that energy savings are exponentially linked to reduced traffic.

D. Benefits The first advantage of the process according to the invention is that it is not linked to the point PA and that it can also be used on the station. However, the architecture of the mobile can possibly prevent the use of the two higher principles because they impose much more processing on the frames, while the lower principle requires only an encapsulation and a reconstitution of multiple frames without filtering on the headers of one or the other level.

In addition, the aggregation principles are fully compatible with the PSP mode which is included in the IEEE 802.11 standard, but can be used alone for real-time exchanges which the PSP mode cannot take into account due to its large latency. In fact, frame aggregation is designed to be complementary to the PSP mode because it is powerful when the PSP mode is weak (high bandwidth or real-time communications). Likewise, the higher the bandwidth, the greater the savings and the lower the latency time.

In order to validate their theoretical analysis, the inventors implemented the process of aggregating frames on a multimedia stream and observed its effects. To achieve such an implementation, they have created tools which allow the streaming of files with parametric packages. A. The mobile device

The mobile used is equipped with two batteries: a main battery which provides power to the mobile, and an additional battery located in the PCMCIA jacket. This design made it possible to accurately measure the consumption of the network interface because the battery of the jacket is reserved for communications, and only discharges when the network interface is used.

At the same time, the inventors were able to assess the increase in power consumption of the central unit involved in the aggregation process by monitoring the discharge of the main battery and comparing it to that which we obtained without any aggregation. We expect this additional load to be lower than the amount of energy saved by our process.

B. The Aggregation Validation Platform The inventors have created an aggregation platform in order to be able to evaluate the effect of frame aggregation on an IEEE 802.11 wireless link. It is made up of two parts: a flow encapsulator and a flow unencapsulator. The encapsulation part sends data from a stream containing packets using several parameters, the IP version, the size of the third-level packet, the load of the packet (allowing space to be reserved so as to simulate the encapsulation service charge), and the length of the radio frame (for the limitation to 2312 bytes). The reconstruction or decapsulation part takes the incoming packets and separates them into the necessary parts exactly as would be done in the actual implementation. Data is then extracted from the packets and sent to the multimedia application which is used to view the flow.

C. The access point

As indicated previously, the packets coming from outside the WLAN network (wired local network) which are delivered by the PA point come from an Ethernet segment using a MTU of 1500 bytes. Therefore, the application must be run on the PA point itself. This is the reason why the inventors have used what is called a "host PA point" which consists of a computer with a wireless adapter acting as a PA point, allowing us to use the validation platform with third level packets sized at 2312 bytes. D. Implementation of tests

Using these components, it was possible to make measurements in both directions of communication (from the mobile to the PA point, and from the PA point to the mobile). These tests are performed using all the possible combinations of aggregation levels, PSP mode and IP version. The stream used is an hour-long audio sample compressed using a raw bit rate of 192 kilobits per second. Each isolated test is run five times and the results are average.

The efficiency of the aggregation is measured by monitoring the state of the two batteries during the experiment, and by comparing them at each level of aggregation. The bit rate chosen is deliberately low because to be sure that the aggregation will work well, even for such low bit rates. Just as the efficiency of the aggregation depends on the bandwidth, good results on streams with low bit rate would mean even better results for high bit rates.

Among all the tests carried out, the inventors have chosen certain results which constitute illustrations representative of the performance of the principle of aggregation.

A. First case: Mobile to the fixed network In this case, the effects of aggregation are monitored during the streaming from the mobile to the fixed network. The graph in Figure 6 shows that the second battery used for communications drops from 100% to 90% after approximately 30 minutes of streaming. The first curve to fall is that of experimentation which does not use aggregation. Then the second curve drops within two minutes (level two aggregation), the third curve drops one minute later (level two + aggregation) and the last curve also drops two minutes later (level three aggregation).

All these experiments have demonstrated the expected result: the higher level aggregation principles are more effective than the lower principles. The savings achieved by each level range from two to five minutes for an hour-long workflow. Two minutes being obtained by level two, then three minutes by level two + and five minutes by level three. B. Second case: Fixed network to mobile

This second case (see Figure 7) represents the performance of the aggregation when it is used without PSP mode during a streaming from the fixed network to the mobile. As the PSP mode is deactivated, several drops in battery level are visible during the experiment. The first fall takes place around 15 minutes and makes it possible to verify that the aggregations of levels two and two + save one minute, and that level three still saves one minute compared to the normal principle. The second fall takes place between the 30 ^th and 40 ^th minutes and indicates savings ranging from two minutes for level two aggregation, and three minutes for level two + up to five minutes for level three. Finally, the last fall takes place in the last quarter of an hour and indicates that for a flow lasting one hour, the level two aggregation saves four minutes, the level two + saves six minutes and the level three saves nine minutes.

As expected, aggregation is a two-way mechanism that allows savings to be made in both directions of communication. The battery life is increased from four to nine minutes for a flow of one hour, which means that by using the same amount of energy, the mobile can communicate up to nine additional minutes using our new process. C. Cost of aggregation

During all the experiments, the situation of the two batteries was monitored, and Figure 8 represents the discharge of the main battery during the first experiment, where the mobile does not restore any flux, so that the energy consumed from the battery is mainly used by the processor and memory. The differences between all the curves are so small that they often cross each other and almost merge.

Thus, the new process proposed by the inventors is based on a difference in design between the wired and wireless local area networks and works well under the conditions of the tests used to validate its effectiveness, although they were far from ideal circumstances. The bit rate chosen was low enough to evaluate a simple high quality audio stream, but as the bit rate increases, the power consumption using aggregation will increase more slowly than that using no aggregation principle.

These savings would allow the battery life to increase by at least about 10% in the worst conditions, and as explained in the above paragraph, the results would be even better with streams at higher bit rates. This process is really interesting because it can also be used in conjunction with the PSP mode of the IEEE 802.11 standard, with fairly good efficiency. And when the PSP mode cannot be used (in particular in the case of high-bit-rate real-time streams from videoconferencing) the mechanism proposed by the invention can be used alone with excellent efficiency.

Thanks to the above validation, the usefulness of implementing the method according to the invention in the NIC card pilots and in the PA points, because the theoretical results have been confirmed by an implementation and measurements on real equipment. Although the highest level of aggregation could possibly consume too much memory to be implemented in the mobile network stack, mobiles will still benefit from this principle for incoming communications.

The method according to the invention therefore proposes a bidirectional aggregation / disaggregation mechanism, preferably applicable in wireless infrastructure networks (in particular IEEE802.i 1) and having different performance levels in terms of energy saving.

This process can be easily implemented and allows an incremental deployment, making it possible to mix the users equipped with the invention and those who are not, thanks to the total integration of the mechanism proposed by the invention in the existing standard. It should be noted that the implementation of the method does not require the presence of an interconnection network or a central point ("backbone"), and can therefore be applied advantageously to local area networks (LAN) or to wireless LAN cells.

Furthermore, the method according to the invention always allows level two aggregation and systematically respects the order of the frames during their processing, without causing any disequencing.

Finally, the inventive method is extensible, in particular by the use of the generic standard IEEE802.il, making it possible to exploit different physical layers at different bit rates. This also allows the application of the method to physical layers not yet defined, as long as they are compatible with the standard used. Of course, the invention is not limited to the embodiments described and shown in the accompanying drawings. Modifications remain possible, in particular from the point of view of the constitution of the various elements or by substitution of technical equivalents, without thereby departing from the scope of protection of the invention.

Claims

1. Method of communication by transmission of frames or data packets, using at least first and second communication networks linked together by at least one interface or access point, said at least two networks accepting lengths or maximum sizes of different frames or packets, method characterized in that it consists in carrying out, at said access point or at each access point (PA), on the one hand, an aggregation of frames (Tl ) for messages passing from the first network (PR) with a lower maximum frame size to the second network (SR) with a higher maximum frame size and, on the other hand, a disaggregation of frames (T2) for messages passing from the second network (SR) with a maximum maximum frame size (T2) to the first network (PR) with a maximum maximum frame size, followed by a reconstruction of the frames (T1) having been the subject of an aggregation operation beforehand .

2. Method according to claim 1, characterized in that the frame aggregation operation (Tl) is preceded by an operation of removing redundant data between the different frames (Tl) of each group of circulating frames (Tl) on the first network (PR) and intended to form a frame (T2) able to circulate on the second network (SR) and in that the operation of reconstitution, in their entirety or not, of the frames (Tl) intended to circulate on the first network (PR) following a disaggregation of frames (T2) circulating on the second network (SR), is preceded by an operation to regenerate the redundant data deleted during the aggregation.

3. Method according to any one of claims 1 and 2, characterized in that the frames (T2) circulating on the second network (SR) and formed by the aggregation of several frames compatible with the first network (PR) contain additional information concerning the structure of the corresponding aggregations.

4. Method according to any one of claims 2 and 3, characterized in that the data deleted during the aggregation and regenerated during the reconstruction consist of identical data at the level of the headers of the frames (Tl) to be aggregated .

5. Method according to any one of claims 1 to 4, characterized in that it consists in aggregating or grouping frames (Tl) of the same type or nature and whose destinations in the second network are identical, without modification , nor alteration of said aggregated frames (T1).

6. Method according to any one of claims 1 to 5, characterized in that it consists in aggregating or grouping frames (Tl) of the same type, originating from the same source and whose destinations in the second network ( SR) are identical, at least part of the header of each frame (T2) to be transmitted on the second network (SR) being common to all the frames (T1) aggregated or encapsulated in this frame (T2).

7. Method according to any one of claims 1 to 6, characterized in that it consists in aggregating or grouping frames (Tl) of the same type, forming part of the same message or of the same data stream, originating from the same source and whose destinations in the second network (SR) are identical, at least part of the header of each frame (T2) to be transmitted on the second network (SR) being common to all the frames ( Tl) aggregated or encapsulated in this frame (T2).

8. Method according to any one of claims 1 to 7, characterized in that the first network (PR) is a wired network, in particular local, and in that the second network (SR) is a wireless or transmission network radio frequency.

9. Method according to any one of claims 1 to 7, characterized in that the two networks (PR and SR) are wireless networks or radio frequency transmission.

10. Method according to any one of claims 1 to 9, characterized in that the transmission in the second network (SR) meets the communication standard known under the designation IEEE 802.11.

11. Method according to claims 6 and 10, characterized in that it consists in aggregating frames (Tl) for which the data of the header level for access control to the support (MAC) are identical.

12. Method according to claims 7 and 10, characterized in that it consists in aggregating frames (T1) for which the data of the header of the network level protocol, for example of the Internet protocol, are identical.

13. Interface or access device intended to provide a link between a first and a second communication network authorizing the circulation of frames or data packets, said two networks accepting maximum lengths or sizes of different frames or packets , characterized in that said device comprises means capable of carrying out, on the one hand, an aggregation of frames (Tl) for messages passing from the first network (PR) with a lower maximum frame size to the second network (SR) at higher maximum frame size, and secondly, a disaggregation of frames (T2) for messages passing from the second network (SR) with higher maximum frame size to the first network (PR) with lower maximum frame size, followed a reconstitution of the frames (Tl) previously aggregated.

14. Device according to claim 13, characterized in that it is suitable for implementing the method according to any one of claims 2 to 12.

15. Device according to any one of claims 13 and 14, characterized in that it is integrated in a mobile or cell phone.

16. Device according to any one of claims 13 and 14, characterized in that it is integrated in an access point unit (PA) ensuring the connection between one or more station (s) of a communication network wireless (SR) and a wired network (PR).

17. Communication system comprising at least three separate communication networks all authorizing the circulation of frames or data packets, one of said networks, called the central network, accepting frames of maximum length or size greater than those accepted by both other networks, and being able to connect these last two networks by means of communication channels or paths being established between at least two interface or access points ensuring respectively a connection between said central network and one of the other two networks, a system characterized in that it implements the communication method according to any one of claims 1 to 12 at each interface or access point (PA).