Patent Application
20060072610
The invention relates to a transmission system, a transmitter for transmitting a data unit to a receiver via a network, and a receiver for processing a data unit received via a network.
The invention also relates to a method used by such a receiver or such a transmitter, and to a computer program using said method.
The invention notably finds its application in the field of transmitting and receiving multimedia data via a network having a limited passband and being liable to errors.
U.S. Pat. No. 6,246,683 B1 describes a method used by a receiver for processing a data unit received via a network. Said receiver 1, described in a functional manner with reference to
As is indicated in
Said receiver 1 described in the above-cited United States patent comprises, at the level of the departure layer L1, means for using a network interface IR, intended to separate the control information IC from the useful information IU within said data unit UDR. The control information IC is subsequently processed in the normal manner by the means for using the network stack 2, while the useful information IU is transmitted to the application layer L7 via said direct connection 3.
Such a method has the advantage that the processing of a data unit UDR received by said receiver 1 in terms of using memory resources is optimized. The fact that the useful information IU is conveyed via said direct connection provides the possibility of avoiding a certain number of memory copies of said information which would necessitate their passage in the network stack 2.
It is an object of the invention to use such a direct connection between a departure layer and an arrival layer of a network stack for another purpose.
A receiver according to the invention as described in the opening paragraph is characterized in that it comprises:
The invention allows the optimization, from a qualitative point of view, of processing a data unit received by said receiver by sending local data available at the level of said departure layer to said arrival layer.
Indeed, in the field of transmitting multimedia data, the networks used are networks having a limited passband and a high error rate, such as, for example, wireless networks. One means of correcting said errors is to send, to the application destination, i.e. to the application layer of the network stack, data which are locally available at the level of the receiver such as, for example, a transmission channel state at the moment of passage of said data unit or a probability that said data unit has errors. A major problem is that said local data must make use of the network stack so as to be transmitted to said application layer and must thus be packeted in a data unit in conformity with the protocols which govern said stack. This is a relatively complex operation necessitating detailed knowledge of said protocols.
The invention therefore proposes a simple solution for sending local data from a departure layer to an arrival layer in a network stack, with the aid of a direct connection between the two layers. This solution is advantageous because it does not require any previous knowledge of the protocols governing said stack.
Said direct connection may be an open connection, for example, by means of drivers or sockets which are placed at the level of the departure layer and the arrival layer of said stack.
It should be noted that the sockets mentioned here are of a particular type, very similar to those used in the field of information security for constructing firewalls.
As such a connection is open for sending a known type of local data from the departure layer to the arrival layer in the network stack, a minimum of control information is necessary to characterize such data. Packeting said local data and said control information in one and the same data structure is certainly necessary but said data structure is much reduced. The means to be used for packeting and subsequent unpacketing said data are thus very simple and consequently the transfer of said data from the departure layer to the arrival layer is realized very rapidly.
It should be noted that such a direct connection may a priori be open between two arbitrary layers of a network stack, particularly in the case of use with drivers. In contrast, if sockets are used, said connection can only be established between a lower layer (the physical layer or the connection layer) and the application layer of said stack.
Another advantage of this solution is that it is independent of the protocols governing the network stack. It is thus valid irrespective of the protocols used and suitable for any type of receiver, provided that it can establish a direct connection between two layers of a network stack. Moreover, such a solution does not affect the actual operation of said stack and consequently does not disturb it.
When the local data generated by the departure layer of the network stack relate to a data unit received by the receiver, the receiver according to the invention also comprises marking means intended to associate said local data with said received data unit, by adding a marker to them.
A possible marker is the relevant received data unit. The advantage of such a marker is that the independence of the solution with respect to the protocols used by the network stack is preserved.
In a first embodiment, said generated local data relate to the state of the channel. Such data may be helpful for the destination application in deciding whether erroneous received data must be corrected or whether retransmission to the transmitter is required.
In a second embodiment, said local data relate to probabilities that the received data are erroneous. Such probabilities may be helpful for the destination application in marking the erroneous received data in order to process them accordingly.
The invention also relates to a transmitter for processing data to be transmitted to a receiver via a network. Said transmitter comprises means for using a network stack and means for establishing a direct connection between a departure layer, for example, an upper layer, and an arrival layer, for example, a lower layer, of said stack.
In a third embodiment, said departure layer provides local data indicating an importance of the data to be transmitted to the receiver via the network. Such data may be advantageously used by a channel encoder for protecting the data to be transmitted in a differential manner. A second advantage is that, by further protecting the most important data to be transmitted, there is a smaller risk that these data are lost and, consequently, the number of retransmissions required by the receiver in the case of erroneous data units is limited. The use of the passband of the network is thus optimized.
These aspects of the invention as well as other, more detailed aspects will be more clearly apparent from the following description of several embodiments of the invention, given by way of non-limiting examples, and with reference to the accompanying drawings in which:
a is a functional diagram of a receiver comprising means for using a network stack and means for direct connection between a departure layer and an arrival layer of said stack, in accordance with the prior art,
b describes the structure of a data unit intended to be transmitted via a network, in accordance with the prior art,
Data DE are transmitted by a source application SAPP of said transmitter EM and then processed by said stack PR′. A data unit UDE is transmitted through the transmission channel of the network R. A data unit UDR is received by the network stack PR of said receiver REC. Received data DR are supplied to a destination application DAPP.
In a first embodiment of the invention, the receiver REC shown in
In this first embodiment, the generating means GENER comprise sub-means MEAS for measuring the state of the channel EC, which means measure a data M and transform it into local data DL describing the state of the channel EC. It concerns, for example, an error rate.
Said local data DL (EC in the example of the state of the transmission channel) are subsequently processed by packeting means PACKET intended to packet said local data in order to render them usable by the arrival layer L7 which will receive them via the direct connection CD. Said packeting means PACKET supply a data structure SDL which is, for example, organized in the way as shown in
These three fields are sufficient when the local data DL to be transmitted are not associated with any data unit UDE transmitted by the transmitter EM and sent to the destination application DAPP by the network stack. This may be the case for the state of the channel EC. One may then consider that the local data received by the destination application DAPP are valid until the next update.
Such a structure SDL also allows simultaneous transmission of several local data of different types to the destination application DAPP by concatenating them within one and the same structure.
In contrast, if the local data concern a particular received data unit UDR, said packeting means PACKET also comprise means MARK for marking said local data structure SDL, intended to mark said structure by means of a field Mk, which is characteristic of said data unit UDR.
For example, the state of the channel EC is a local data which varies in time and, by virtue thereof, the validity of a measure of the state of the channel is generally limited to the transmission of a data or a series of data. In the case where a marking of said structure SDL is necessary, three supplementary fields are used, as is shown in
A marked data structure SDLM is then supplied to the direct connection CD. The marked data structure SDLM is subsequently sent to the arrival layer, in this case the application layer L7 via said direct connection CD. Said layer L7 comprises retrieving means RETRIEV, shown in
Said retrieving means RETRIEV are very simple in the case where the local data to be retrieved are independent, i.e. where they are not associated with any received data DR from the network stack PR. It is then sufficient to know the organization of the fields of the local data structure SDL so as to be able to read it. The retrieving means RETRIEV are thus essentially reduced to sub-means READ for reading said data structure SDL, intended to identify the relevant local data DL in the structure SDL.
In contrast, in the case where the local data to be retrieved relate to a received data unit UDR, said read sub-means READ do not only isolate the local data DL but also the marker Mk. Moreover, said retrieving means RETRIEV also comprise associating sub-means ASSOC intended to search with which received data DR said local data are associated. Such sub-means try, for example, to find a common data in the marker Mk and in the received data DR.
The choice of the marker Mk may be related to control information IC contained in the decoded data unit UDD and characteristic of said data unit UDD such as, for example, a sequence number. However, such a choice would require knowledge of the protocols used by the network stack PR. If, in contrast, one chooses the marker Mk to be equal to the decoded data unit UDD in question, no knowledge of the protocols is required. Indeed, since said marker contains a copy of said received data DR, the association of local data with decoded data will be evident. In this case, the associating sub-means ASSOC of the application layer L7 easily associate the local data structure SDLM with the corresponding received data DR by means of a simple correlation measure.
In the preferred embodiment shown in
At the level of the physical layer L1, a channel decoder CDEC supplies, for a received data unit UDR, a real signal which is constituted by a succession of real data. Said signal may be processed in two different manners:
In the case of a transmission network having a high error rate, the knowledge of such probabilities is very advantageous at all levels of the receiver and particularly at the level of the destination application DAPP. Indeed, such knowledge allows a more precise interpretation of a received data DR and facilitates a possible correction of errors. In contrast, from a point of view of the network stack PR, these probabilities cannot form part of the decoded data unit UDD transmitted by the physical layer L′1 to the upper layers of the network stack. Indeed, a layer protocol of the network stack PR will not accept a flexible decoded data unit UDDS comprising a number of bits which is different from that of the transmitted data unit which has traversed the corresponding layer of the network stack PR. Said probabilities must thus be considered as local data generated by the generating means GENER situated at the level of the physical layer L1. Said generating means GENER comprise thresholding sub-means THRES and quantizing sub-means QUANT using techniques which are well known to those skilled in the art.
The packeting means PACKET subsequently supply a local data structure SDL which is also shown in
As has been explained hereinbefore, the marker Mk used for marking the local data structure SDL containing the flexible bits associated with the hard bits of the received data DR is the decoded data unit itself in the preferred embodiment. An advantage of such a solution is that it is independent of protocols used and that it facilitates the assignment of local data to received data.
Nevertheless, it should be noted that, in the preferred embodiment, the invention is not completely independent of the protocols for the network stack PR. However, only the knowledge of a type of control information used in the majority of network stack models is required. It concerns control information used by at least one layer protocol of the network stack, namely the UDP protocol of the transport layer L4, referred to as “checksum”. Said checksum has a value which is equal to the sum of the bits forming a data unit transmitted during its passage in the corresponding layer L3 of the network stack PR′ and before it is sent on the network. At the receiving end, the corresponding layer of the network stack PR computes a new sum from the decoded data unit UDD. If it obtains a value which is identical to the checksum figuring in the control information of the data unit, said data unit UDD is declared valid by said layer. In the opposite case, it is rejected.
It is precisely for improving the performance of the receiver by correcting transmission errors that one wishes to transmit flexible data to the destination application DAPP. Therefore, the erroneous decoded data units should not be blocked at the level of the network stack PR2 so as to give them an opportunity to be corrected by the destination application with the aid of local data DL supplied via the direct connection.
Consequently, in the preferred embodiment, the computation of the checksum UDP is inhibited, such that the decoded data units UDD which have been declared erroneous in accordance with this criterion are not rejected. The preferred embodiment of the invention which has been described hereinbefore thus punctually intervenes in the operation of the network stack and for this purpose requires the knowledge of a single type of widely used control information.
A third embodiment of the invention, illustrated in
The generating means GENER′ also comprise weighting sub-means WEIGHT intended to weight the importance of the types of data discriminated on the basis of the a priori knowledge CAP2. In the previous example of an encoder of a sequence of images of the MEPG video type, one may consider that the data of the motion type MV are more important than the data of the texture type TEX. Indeed, an encoding scheme of the MPEG type realizes a motion compensation of a current image with respect to a preceding image. Only a difference of texture between the current image and the preceding image which is motion-compensated is transmitted to the destination application DAPP. Consequently, without these motion data, the destination application DAPP, namely the source decoder, would not be able to reconstruct an acceptable current image from texture data only, whereas the contrary is possible. Consequently, by supplying this type of local data to the channel encoder CENC, it is given the means for performing an unequal error protection of the data adapted to the types of transmitted data.
The application layer L′7 also comprises packeting means PACKET′ intended to structure the local data DL′ supplied by said generating means GENER′.
The physical layer comprises retrieving means RETRIEV′, intended to retrieve the local data DL′ within said structure SDLM′. Said retrieving means RETRIEV comprise sub-means READ′ for reading the structure SDLM′, intended to extract the local data DL′ and the marker Mk′, and sub-means ASSOC′ for associating said marker Mk′ with a transmitted data unit UDE, intended to retrieve the marker Mk′ within said transmitted data unit UDE. In a manner analogous to the embodiments of the invention described hereinbefore, this operation will be very simple in the case where said marker is chosen to be equal to the transmitted data.
The invention is not limited to the embodiments which have been described by way of example. These embodiments may be modified or improved without departing from the scope of the invention.
The description above with reference to FIGS. 1 to 9 illustrates rather than limits the invention. It will be evident that there are other alternatives which can be used without departing from the scope of the appendant claims.
There are numerous ways of implementing the functions described by means of software. In this respect, FIGS. 1 to 9 are very diagrammatic, each Figure only representing an embodiment. Although a Figure shows different functions in the form of separate blocks, this does not exclude that a single piece of software performs several functions. This neither excludes that a function can be performed by a software assembly.
It is possible to implement these functions by means of a receiver circuit comprising one or several suitably programmed processors. A set of instructions stored in a programming memory may cause the circuit to perform different operations described hereinbefore with reference to FIGS. 1 to 9. The set of instructions is, for example, loaded into the programming memory by reading a common data carrier, for example, a CD-ROM. In another embodiment, the reading may also be realized by means of a communication network such as the Internet. In this case, a service provider puts the set of instructions at the disposal of those interested.
Any reference sign between parentheses in a claim should not be construed as limiting the claim. Use of the verb “comprise” and its conjugations does not exclude the presence of elements or steps other than those stated in a claim. Use of the article “a” or “an” preceding an element or a step does not exclude the presence of a plurality of such elements or steps.
| Number | Date | Country | Kind |
|---|---|---|---|
| 02/14035 | Nov 2002 | FR | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/IB03/04832 | 10/29/2003 | WO | 5/6/2005 |
