Multipoint voice-communication system in quasi full-duplex mode

TECHNICAL AREA

The present invention relates generally to a voice communication system for connecting a plurality of users to each other.

STATE OF THE PRIOR ART

Voice communication systems for a group of users are known in the professional or rescue world. For example the PMR system (acronym of<i>Professional Mobile Radio</i> or <i>Private Mobile Radio</i>) operates around 446 MHz and offers “walkie-talkie” type services. The narrow system band (12.5 kHz) does not allow more than one user to speak at a time. The communication is of the half-duplex type with a "Push to Talk" function, switching from reception mode to transmission mode by pressing a button on a radio terminal.

A digital version of the PMR system, called DMR (<i>Digital private Mobile Radio</i>) is being normalized. A description can be found in particular in the documents ETSI TS 102 36 and ETSI TS 102 398 available under the site<u>www.etsi.org</u>. This system also uses narrowband channels (12.5 kHz) and allows half-duplex communications.

In general, most of the voice communication systems known in the state of the art use few transmission resources (for example low bandwidth) so that full-duplex communication is hardly possible. In addition, most of these systems only allow point-to-point communication, in other words, at a given moment, only two users in the group can communicate with each other. Finally, they have the disadvantage of requiring manual operation on a radio terminal (press a button) each time a user in the group wants to speak to another user. This manual operation can be awkward, dangerous or even impossible during certain rescue missions.

The object of the present invention is therefore to provide a voice communication system overcoming the aforementioned drawbacks.

STATEMENT OF THE INVENTION

The present invention is defined by a method of voice communication between a plurality of users equipped with radio terminals, called user nodes, in which the user nodes access a common transmission resource, in a manner controlled by a control node, the access to said resource being organized in time frames, each time frame comprising:<ul id="ul0001" list-style="dash" compact="compact"><li>a first period, called the polling period, during which the user nodes having a voice packet to transmit, send a request for access to the control node;</li><li>a second period, called the allocation period, during which the control node allocates transmission frames to all or part of the set of user nodes having transmitted an access request in the previous step, each transmission frame being made up of a plurality of transmission intervals, said frames being interleaved in time;</li><li>a third period, called the transmission period, during which each user node to which a transmission frame has been allocated in the preceding step, transmits said voice packet over the transmission frame which has been allocated to it.</li></ul>

The first period can be divided into elementary polling intervals, each user node having its own elementary polling interval and, if it has a voice packet to transmit, sends its access request during this interval.

Advantageously, the elementary polling intervals are initially scheduled from MAC identifiers of the user nodes, said scheduling being known to the control node and to the user nodes.

This scheduling can be the subject of a permutation at each time frame, the permutation law then being known to the control node and to the user nodes.

The access request may include a priority level of access to said resource, which may itself be a function of the volume of the user's voice.

Each user node advantageously comprises a transmission buffer and transmits an access request within its elementary polling interval if a voice packet is present in said buffer.

Alternatively, each user node comprises a transmission buffer and systematically transmits an access request within its elementary scanning interval, the degree of priority being zero in the absence of voice packet within said buffer and being non-zero otherwise .

The control node preferably broadcasts to all of the user nodes, during the allocation period, an allocation message indicating to each user node having transmitted an access request during the preceding scanning period, the frame rank. allocated to it within the transmission period.

According to a first variant, the control node allocates transmission frames only up to a predetermined number of user nodes which have made an access request.

According to a second variant, the control node determines a maximum number, <i>NOT_s</i>, of user nodes that can be the subject of a transmission frame allocation, said maximum number being determined by the largest integer <i>NOT_s</i> checking: <maths id="math0001" num=""><math display="block"><mi mathvariant="italic">PP</mi><mo mathvariant="italic">+</mo><mi mathvariant="italic">AP</mi><mo mathvariant="italic">+</mo><msub><mi mathvariant="italic">NOT</mi><mi mathvariant="italic">weft</mi></msub><mstyle displaystyle="true"><munderover><mo>∑</mo><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><msub><mi>NOT</mi><mi>s</mi></msub></munderover></mstyle><msub><mi>T</mi><mrow><mi>f</mi><mo>,</mo><mi>i</mi></mrow></msub><mo>≤</mo><msub><mi>T</mi><mi>F</mi></msub></math><img file="EP2472914A1_D0001.tif" /></maths> or <i>T_F</i> is the duration of the time frame, <i>T_{ƒ, i}</i> that of the user's transmission frame <i>i, PP</i> is the length of the scanning period, <i>AP</i> is the duration of the allocation period, <i>NOT_weft</i> is the number of interleaved transmission frames, an interlaced transmission frame being constituted by a temporal interleaving of the transmission frames of the user nodes.

Advantageously, during the transmission period, each user node to which a transmission frame has been allocated during the preceding allocation step, broadcasts its voice packet to all of the user nodes.

The control node can also be a user node.

Prior to any communication, an initialization step is preferably carried out comprising:<ul id="ul0002" list-style="dash" compact="compact"><li>a step of automatically discovering user nodes by exchanging their respective MAC addresses;</li><li>a step of electing a control node from among the user nodes;</li><li>a step of synchronizing the user nodes with the control node thus elected.</li></ul>

The control node can associate with each MAC address of a user node, a MAC address identifier, and then broadcast to all of the user nodes a message containing the MAC address of each user node and the identifier MAC address thus associated.

BRIEF DESCRIPTION OF THE DRAWINGS

Other characteristics and advantages of the invention will appear on reading a preferred embodiment of the invention, made with reference to the attached figures among which:<ul id="ul0003" list-style="none" compact="compact"><li>The <figref idref="f0001">Fig. 1</figref> schematically illustrates an initialization phase prior to the implementation of the voice communication method according to an embodiment of the invention;</li><li>The <figref idref="f0002">Fig. 2</figref> schematically illustrates the method of voice communication according to an embodiment of the invention;</li><li>The <figref idref="f0003">Fig. 3</figref> schematically represents the structure of a time frame used in the context of the present invention;</li><li>The <figref idref="f0004">Fig. 4</figref> schematically illustrates the implementation of said method by the network control node;</li><li>The <figref idref="f0005">Fig. 5</figref> schematically illustrates the implementation of said method by a user node of the network;</li><li>The <figref idref="f0006">Fig. 6A</figref> shows an example of the structure of a packet used in a first step of said voice communication method;</li><li>The <figref idref="f0006">Fig. 6B</figref> shows an example of the structure of a packet used in a second step of said voice communication method;</li><li>The <figref idref="f0006">Fig. 6C</figref> shows an example of the structure of a packet used in a third step of said voice communication method.</li></ul>

DETAILED PRESENTATION OF PARTICULAR EMBODIMENTS

We will consider below a network of radio terminals, also referred to below as user nodes. The network also includes a node responsible for coordinating the network, called a control node. The control node can be a particular user node of the network. This will be particularly the case in a cooperative type network, where all the nodes have the same structure and where each node can be called upon to play the role of control node. Alternatively, the control node can be a dedicated node, exclusively responsible for coordinating the network. This will be the case in particular in a cellular type network where the control node will then be a base station.

We will assume hereinafter, by way of illustration and without prejudice to generalization, that the network is of the cooperative type and that the control node is also a user node of the network.

We will also assume that communications over the network use a narrow transmission band and that users access this common resource by time division multiple access or TDMA (<i>Time Division Multiple Access</i>). In other words, the transmission resources of the different users are transmission intervals or TTI (<i>Time Transmission</i> Interval) during which they can transmit in the band in question.

The implementation of the voice communication method according to the invention assumes that an initialization of the network has been carried out beforehand. It should however be emphasized that this initialization step is not part of the invention itself.

The <figref idref="f0001">Fig. 1</figref> illustrates said initialization step in the case of a cooperative network.

This initialization step can be launched when the radio terminal of any user is powered up.

In a first phase, 110, an automatic discovery of the network is carried out by each of the nodes. At this stage, the different nodes exchange their MAC addresses and, where appropriate, some of their characteristics. These exchanges are carried out by packet transmission in asynchronous mode. In the event of a collision, the packets are retransmitted until all the nodes have recognized each other.

In step 120, the control node is elected according to a predetermined rule. The takeover by this node advantageously gives rise to the broadcasting of a message to all the nodes of the network indicating the identity of the control node, typically its MAC address.

In step 130, the user nodes are synchronized with the control node. To do this, the control node broadcasts for example a synchronization packet to all the nodes of the network. The user nodes synchronize using this packet to the clock of the control node.

At the end of the initialization step, each node knows the size of the network (number of users) and the MAC addresses of the other nodes. To simplify subsequent exchanges, a node identifier (denoted MAC-Id) coded on a few bits (for example a quadruplet) is advantageously associated in a one-to-one manner with each MAC address. This association can be carried out centrally by the control node and declared to the user nodes by a packet containing the MAC address of the node as well as its MAC-Id identifier.

Different variants of the initialization step could be envisaged by those skilled in the art without departing from the scope of the present invention. In particular, it is conceivable that a particular user node can be configured as a control node from a configuration switch provided on the radio terminal. Synchronization can be carried out by grouping all the terminals in one place and manually initiating this synchronization. In the case of a dedicated control node, it can play the role of docking station (<i>docking station</i>)<i>,</i> the various terminals then being connected to this docking station and synchronizing with it.

The <figref idref="f0002">Fig. 2</figref> represents the main steps of the voice communication method according to an embodiment of the invention.

Access to the common transmission resource is organized by dividing the time into consecutive time frames, each frame comprising a first period, called the scanning period, a second period called the allocation period, and a third period called the transmission period.

In step 210, during the Polling Period (PP), the control node receives requests for access to the transmission resource from the various user nodes. More precisely, the scanning period is itself divided into elementary intervals, called elementary scanning intervals, each user node having its own interval for transmitting a request for access to the resource. The order of these intervals within the scanning period is known to the user nodes and to the control node by virtue of a predetermined convention. For example, the order of these intervals may be the ascending or descending order of the MAC-Id identifiers assigned during the initialization step. This order can however be subsequently modified by the control node as explained below.

In step 220, the control node allocates transmission frames to the user nodes having requested access in step 210. In general, the allocation may not satisfy the <i>NOT_r</i> user nodes having transmitted an access request, a plurality <i>NOT_s</i>, with <i>NOT_s</i> ≤ <i>NOT_r</i>, which can be served. The number<i>NOT_s</i>, can be fixed or calculated by the control node as explained below.

A transmission frame consists of a plurality of transmission intervals, these intervals being obtained by dividing the transmission period. The transmission frames allocated to different user nodes are interleaved.

The allocation of transmission frames can take into account the priority levels of the different users. After performing this allocation, the control node broadcasts an allocation message to all of the user nodes during the allocation period (PA).

In step 230, the user nodes to which transmission frames have been allocated transmit their respective voice packets by means of said frames during the transmission period (TP).

The voice communication method continues by returning to the first step.

The <figref idref="f0003">Fig. 3</figref> schematically represents the structure of a time frame used in the context of the present invention.

As indicated above, the time frame comprises a polling period (PP), 310, an allocation period, 320, and a transmission period 330.

The polling period, 310, is divided into as many elementary intervals, 311, as there are user nodes. During the elementary interval which is assigned to it, a user node can transmit a request for access to the control node.

The allocation period (AP), 320, is used by the control node to broadcast an allocation message to all of the user nodes.

The transmission period (TP), 330, includes a plurality <i>NOT_weft</i> of interlaced transmission frames, each frame consisting of a plurality <i>NOT_s</i> transmission intervals, one for each user node. We assumed here that<i>NOT_s</i> = 2, that is, only two users were allowed to speak during the transmission period. The frames are classified by rank, the rank of a frame being the rank of the first transmission interval which composes it. Thus, in the illustrated case, the first frame consists of the transmission intervals, represented in clear, of odd rank and the second frame consists of the transmission intervals, represented in gray, of even rank.

Although a single user can transmit during a given transmission interval, it will be understood that the interlacing of the frames results in the perception of two “simultaneous” half-duplex communications, in other words a full-duplex communication. The degree of fragmentation of the frames is chosen sufficiently high so that the reproduction of the voice does not present any blanks.

Generally, if <i>T_F</i> is the total duration of the time frame, <i>Tƒ_{, i}</i> the duration of the voice packet transmitted by user i, the maximum number <maths id="math0002" num=""><math display="inline"><msubsup><mi>NOT</mi><mi>s</mi><mi>max</mi></msubsup></math><img file="EP2472914A1_D0002.tif" /></maths> of "simultaneous" half-duplex communications is given by the largest integer <i>NOT_s</i> checking the relation: <maths id="math0003" num="(1)"><math display="block"><mi mathvariant="italic">PP</mi><mo mathvariant="italic">+</mo><mi mathvariant="italic">AP</mi><mo mathvariant="italic">+</mo><msub><mi mathvariant="italic">NOT</mi><mi mathvariant="italic">weft</mi></msub><mstyle displaystyle="true"><munderover><mo>∑</mo><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><msub><mi>NOT</mi><mi>s</mi></msub></munderover></mstyle><msub><mi>T</mi><mrow><mi>f</mi><mo>,</mo><mi>i</mi></mrow></msub><mo>≤</mo><msub><mi>T</mi><mi>F</mi></msub></math><img file="EP2472914A1_D0003.tif" /></maths>Or : <maths id="math0004" num="(2)"><math display="block"><msub><mi>T</mi><mrow><mi>f</mi><mo>,</mo><mi>i</mi></mrow></msub><mo>=</mo><mfrac><mrow><msub><mi>η</mi><mi>i</mi></msub><mo>⋅</mo><msub><mi>k</mi><mi>i</mi></msub><mo>⋅</mo><msub><mi>ρ</mi><mi>i</mi></msub></mrow><msub><mi>γ</mi><mi>i</mi></msub></mfrac><mo>+</mo><mi>α</mi></math><img file="EP2472914A1_D0004.tif" /></maths>and or :<ul id="ul0004" list-style="none"><li><i>^PP</i> is the length of the scanning period, <i>^AP</i> is the duration of the allowance period; <i>ηi</i> is the time duration of each elementary voice interval, <i>k_i</i> is the number of elementary voice intervals per user, <i>ρ_i</i> is the compression rate of the voice,</li><li><i>γ_i</i> is the bit rate used by the user node <i>ⁱ</i> ,</li><li>and <i>α</i> is a time margin necessary in practice for the transmission of each voice packet. The voice stream of each user is fragmented into a plurality of elementary frames and each voice packet transmitted during a transmission interval contains several elementary frames.</li></ul>

After demodulation and decoding of the voice packets, the following relation must be satisfied so that the voice signal does not present blanks during the restitution: <maths id="math0005" num="(3)"><math display="block"><msub><mi>η</mi><mi>i</mi></msub><mo>⋅</mo><msub><mi>k</mi><mi>i</mi></msub><mo>⋅</mo><msub><mi mathvariant="italic">NOT</mi><mi mathvariant="italic">weft</mi></msub><mo>≥</mo><msub><mi>T</mi><mi>F</mi></msub></math><img file="EP2472914A1_D0005.tif" /></maths>

The relation (3) means that after demodulation and decoding of the voice packets, the duration of the restitution of the vocal signal of each user cannot be less than the length <i>T_F</i> of a time frame.

The <figref idref="f0004">Fig. 4</figref> illustrates the implementation of the voice communication method by the control node.

In step 410, at the start of the polling period, the control node goes from the standby state to the active state.

In step 420, during the entire period of the scanning period, the control node collects the access requests transmitted by the various user nodes. These requests can indicate the priority of the access requested.

Step 430 is optional and relates to the case where the control node is also a user node, that is to say a radio terminal of one of the users of the network. In this case, the control node checks whether a voice packet is present in its transmission buffer. If this is the case, the access request of the control node is taken into account locally at 435. If this is not the case, no access request is generated.

In step 440, the control node allocates the transmission frames. This allocation may take into account the degrees of priority, when these are specified in the access requests. Alternatively or alternatively, the allocation may take into account a degree of priority implicitly provided by type of radio terminal (coded in the MAC address). When a plurality of users has the same degree of priority, the allocation of transmission frames is done randomly within said plurality. At the end of this step, the user nodes having requested it, or only some of them (up to<i>NOT_s</i>) are allocated a transmission frame.

In step 450, the control node broadcasts to all of the user nodes an allocation message indicating to each user node concerned the rank of the transmission frame which has been allocated to it. More precisely, this message advantageously contains a correspondence between the identifiers of the user nodes (MAC-Ids) and the ranks of the transmission frames. It will be understood that if the control node is also a user node the allocation message does not contain the allocation information for the control node but that this information is stored locally.

In step 460, the control node switches from the active state to the standby state until the start of the next scanning period. However, when the control node is also a user node, the standby state is interrupted to transmit its own voice packet (or to receive voice packets from other users). The control node is then in the active state during the transmission of the packet and immediately returns to the standby state until the start of the next scanning period.

The <figref idref="f0005">Fig. 5</figref> illustrates the implementation of the voice communication method by a user node (excluding the control node).

At the start of the polling period, the user node switches from standby to active state in 510. It determines in 520 whether a voice packet is to be transmitted, in other words if a voice packet is present in its buffer. transmission.

In the absence of a packet to be transmitted, it returns to the standby state in 540 until the start of the allocation period. On the other hand, if a voice packet is to be transmitted, the user node transmits in step 530 its access request during the elementary polling interval which is assigned to it. At the start of the communication, this assignment is preferably a predetermined function of the MAC-Id. However, this assignment can be changed as explained below.

The access request advantageously comprises a degree of priority specified by the user. For example, the priority may be an increasing function of the volume of the voice. Once its access request has been transmitted, the user node returns to standby in 540, until the start of the allocation period.

In 550, the user node changes from the standby state to the active state at the start of the allocation period. It receives the allocation message broadcast by the control node. It determines from this message whether a transmission frame has been allocated to it for the current time frame and, if so, the rank of this transmission frame.

In step 560, the user node transmits its voice packet during the transmission frame which has been allocated to it. More precisely, it fragments this packet and transmits it during the transmission intervals constituting the frame in question. It simultaneously receives voice packets from others (<i>NOT_s</i>-1) user nodes. At the end of the last transmission frame, the user node goes into standby mode in 570, until the start of the next polling period.

As already indicated above, the allocation, or more precisely the scheduling (<i>scheduling</i>) initial elementary polling intervals are preferably performed based on MAC identifiers (MAC-Id). This scheduling can be constant for the entire duration of the call.

According to an advantageous alternative embodiment, the scheduling of the elementary polling intervals will be swapped at each polling interval to statistically reduce the latency times of the voice packets.

A fixed scheduling has the effect of favoring the user nodes to which the polling intervals closest to the start of the allocation period have been assigned. Thus, a voice packet which would be stored in the transmission buffer of a user node during the polling period will be taken into account if its elementary polling interval is the last of said period when it will not be. if its elementary scanning interval will be the first of this period. In this second case, the allocation request will not be issued during the current polling period but only during the next polling period. In other words, the latency of the voice packet will be increased by at least the duration <i>AP</i>+<i>TP</i> (duration of the allocation period plus duration of the transmission period), which can be penalizing.

The aforementioned variant embodiment switches the scheduling of the elementary scanning intervals. The permutation law can be egalitarian, such as a circular permutation at each time frame, or else favor certain priority user nodes.

The permutation law is preferably known to all the user nodes. Thus, each user node knowing the initial scheduling and the permutation law is able to determine, at each time frame, the elementary scanning interval which is assigned to it. In the event of a variation in the number of user nodes, the control node carries out a new initialization of the scheduling on the basis of the MAC identifiers (Mac-Id) of the nodes present in the network.

The <figref idref="f0006">Fig. 6A</figref> represents the format of an access request packet.

The packet comprises a first part in which the MAC identifier (MAC-Id) of the user node at the origin of the request is stored, a second part containing the MAC identifier of the control node, recipient of the request, and, where appropriate, a third part containing the priority level of access.

According to a particular exemplary embodiment, all the user nodes, including those whose transmission buffer is empty, issue an access request during a scanning period. However, when the transmission buffer is empty, the degree of priority is conventionally set to 0 and the request is not taken into account. This allows the control node to distinguish between an access request packet that has not reached its destination (fading of the transmission channel for example) and an absence of request from a user node.

The <figref idref="f0006">Fig. 6B</figref> represents the format of an allocation message packet.

It is recalled that this message is transmitted by the control node to signify to the user nodes the transmission frames which are allocated to them.

The packet includes a header containing the MAC identifier of the control node, at the origin of the message, a plurality of MAC identifiers whose order indicates the ranks of the transmission frames which are respectively allocated to them: the first frame is allocated to the user node identified by MAC-Idl, the second to that identified by MAC-Id2, and so on.

If necessary, the packet can also contain synchronization information, indicating precisely the start of the transmission period. By default, the start of the transmission period coincides with the end of the allocation period.

The <figref idref="f0006">Fig. 6C</figref> schematically represents the format of a voice packet.

This packet includes a header containing the identifier of the user node at the origin of the packet, as well as the payload itself. If necessary, as indicated in the figure, the packet can include synchronization information indicating when the voice packet was generated, so as to allow time alignment during its restitution.

Generally, a user node transmits its voice packet to all the user nodes of the network, so that each node can restore all of the communications in progress. According to a variant, each node can select the user or users that it wishes to listen to. It then suffices to filter the voice packets whose headers contain the MAC-Id identifiers of the selected users.