WO2003051068A2

WO2003051068A2 - Collision management in a radio communication system

Info

Publication number: WO2003051068A2
Application number: PCT/NZ2002/000274
Authority: WO
Inventors: Clive Douglas Horn; Ian Douglas Royds
Original assignee: Tait Electronics Limited
Priority date: 2001-12-13
Filing date: 2002-12-13
Publication date: 2003-06-19
Also published as: AU2002366677A8; WO2003051068A3; AU2002366677A1

Abstract

An improved system for resolution of collisions in peer to peer mobile communication systems, particularly systems that implement DIIS. A mobile that has control of a channel for transmission purposes releases it on detecting a possible collision on the channel. A mobile unit not in control of the channel may warn the controlling unit on detecting a possible collision on the channel. The collision may be detected in various ways, in any domain, including detection of an announcement on the channel by another unit and detection of non-decodable activity on the channel.

Description

COLLISION MANAGEMENT IN A RADIO COMMUNICATION SYSTEM

FIELD OF THE INVENTION

This invention relates to systems for managing collisions in mobile communications, in particular but not only to systems for digital radio networks such as DIIS (Digital Interchange of Information and Signalling). DIIS is a digital radio standard intended for use in mobile radio communications.

BACKGROUND TO THE INVENTION

Communication between mobile stations (MSs) in a DIIS system can happen either directly, or via a repeater or base station. If the MSs are communicating directly, or via a repeater that simply retransmits the MS signalling without altering its content, then the MSs are performing peer-to-peer operation. If all communication between MSs is governed either by a repeater or a base station (RE/BS), such that there is a master-slave relationship between the RE/BS and the MSs, then the MSs are said to be performing centralised operation.

In a DIIS system, when a voice call is to be established, or a significant amount of data is to be transferred, a connection is established between the parties involved in the transaction. To establish a connection across the air interface, a signalling master reserves the radio channel. In peer-to-peer operation the signalling master will be one of the MSs involved in the connection, and in centralised operation, it will be the RE/BS. In both cases, it is the responsibility of the signalling master to regularly transmit a Channel Status (CS) message, according to the various frame structures being defined in the standard.

The DIIS protocol defines three different states of the physical channel, being the idle, reserved, and payload domains. The current domain is announced in the CS messages, except in the idle domain, in which there might be no signalling master making CS announcements. If the channel is in the idle domain, an MS can attempt to use the channel for a connection. If successful, the channel enters the reserved domain. In the reserved domain, the signalling master makes regular CS announcements, stating who should be involved in the connection. The signalling master might also announce which of the MS population can access the channel using random access. When there is application data (whether voice or some other data) to be transferred, the signalling master announces that the channel is in the payload domain, and most slots in the frame structure are used for the transmission of the payload data.

SUMMARY OF INVENTION

It is an object of the present invention to provide an improved system for resolution of collisions in mobile communications under DIIS, or at least to provide an alternative to existing systems.

In one aspect the invention may be said to consist in a method of operating a mobile unit in a peer to peer radio communication system comprising: announcing control of a channel in the system, transmitting on the channel to another unit, detecting a possible collision on the channel, and releasing the channel. The collision may be detected in various ways, in any domain, including detection of an announcement on the channel by another unit and detection of non-decodable activity on the channel.

In another aspect the invention may be said to consist in a method of operating a mobile unit in a peer to peer radio communication system comprising: receiving an announcement of control by another unit of a channel in the system, receiving a transmission from the other unit on the channel in accord with the control, detecting a possible collision on the channel, and transmitting a warning of the collision to the other unit. The collision may be detected in various ways, in any domain, including detection of an announcement on the channel by another unit and detection of non-decodable activity on the channel.

In further aspects the invention also consists in a mobile unit that is capable of operating according to any of the aspects above and to a communication system that includes or enables such mobile units.

The invention further consists in any alternative combination of features that are indicated in the specification or shown in the drawings. All equivalents of these features are included whether or not explicitly mentioned.

In a preferred embodiment, if a signalling master decodes a number of CS announcements by another signalling master within a certain period of time, then it will cease being a signalling master itself. If a signalling master is announcing that the channel is in the reserved domain, and it detects more than a certain number of slots with non-decodable activity in them within a certain period of time, then it will cease being a signalling master.

If a signalling master is announcing that the channel is in the payload domain, and it detects non-decodable activity in one or both slots of a number of Tl opportunity and RS opportunities consecutively, then it will cease being a signalling master.

If an MS that is a receiver in a call decodes a number of CS announcements by an mtrading signalling master other than its current signalling master within a certain period of time, then it will send signalling to its current signalling master to inform it that the channel might be no longer usable. In the reserved domain, this signalling will be sent in the random access slots, and in the payload domain, it will be sent in Tl or RS slots.

If an MS that is a receiver on a channel that it believes to be in the payload domain detects more than a certain number of slots with non-decodable activity in them within a certain period of time, then it will send signalling to its current signalling master to inform it that the channel might be no longer usable.

If an MS that is a receiver on a channel detects more than a certain number of consecutive CS announcements with non-decodable activity in them within a certain period of time, then it will send signalling to its current signalling master to inform it that the channel might be no longer usable.

If an MS that is a receiver on a channel that it believes to be in the payload domain detects non-decodable activity in one or both slots of a number of Tl and RS opportunities consecutively, then it will send signalling to its current signalling master to inform it that the channel might be no longer usable.

LIST OF FIGURES Preferred embodiments of the invention will be described with respect to the accompanying drawings, of which:

Figure 1 indicates a peer-to-peer direct collision,

Figure 2 indicates a signalling master intruding on members of a group call,

Figure 3 indicates a peer-to-peer repeater collision, Figure 4 indicates behaviour of a signalling master in the reserved domain,

Figure 5 indicates behaviour of a signalling master in the payload domain,

Figure 6 indicates behaviour of a receiving MS in the reserved domain,

Figure 7 indicates behaviour of a receiving MS in the payload domain,

Figure 8 indicates behaviour of a receiving MS indicating a collision to its signalling master,

Figure 9 indicates a signalling master hearing an intruder in the reserved domain,

Figure 10 indicates receivers hearing an intruder in the reserved domain,

Figure 11 indicates failure to receive a payload header,

Figure 12 indicates a receiver hearing an intruder in the payload domain,

Figure 13 indicates a collision between two signalling masters in the payload domain,

Figure 14 indicates a payload receiver hearing an intruder also in the payload domain,

Figure 15 indicates two members of a group call transmitting simultaneous payload headers,

Figure 16 indicates timing of an intruder CS announcement relative to a Tl or RS opportunity, and

Figure 17 shows a typical mobile unit including power control components.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

Referring to the drawings it will be appreciated that the invention may be implemented in a variety of ways in a variety of communication systems. An implementation in a DIIS system is described by way of example only. Implementations involving both single and multiple channels may also be constructed. Details of the mobile units and iirfrastructure that support these systems will be known to a skilled reader and need not be given here.

When a DIIS MS is first turned on, it will listen for CS announcements, beacon announcements by a repeater or base station (yet to be defined in the standard), or for any other channel activity. If it hears channel activity, it will, according to the requirements of the proposed standard, obtain and maintain timeslot timing. If it hears a CS announcement, it will immediately be able to determine the current channel domain. Otherwise, after a certain amount of time, the MS can assume that the channel is in the idle domain.

In centralised operation, in which the RE/BS is always the signalling master, no collision between two signalling masters can occur.

In peer-to-peer operation (either direct or via a repeater), collisions can occur for the following reasons:

• Failure to receive a payload header during a change of signalling master: When a signalling master is holding the channel in the reserved domain, another MS (involved in the call) that wishes to transmit speech or data announces its intention by transmitting a payload header. The new signalling master will transmit its payload in the old signalling master's RA slots.

• Simultaneous attempts to talk or transmit data: If both parties in an individual call, or more than one party in a group call, simultaneously attempt to become signalling master, their payload headers might collide. This type of collision can also occur if the payload headers are transmitted in adjacent slots, as an MS can not listen to the slot immediately preceding its own transmission. An equivalent collision can occur if an MS fails to receive another MS's payload header (and potentially a few slots of payload) due to propagation conditions, and subsequently transmits its own payload header. In all these cases, neither party would be aware of the collision.

• False assessment of the channel domain: An MS listens to the channel to deteraiine its state before becorning signalling master. If it misses a number of CS transmissions from an existing signalling master (due to shadowing, for instance), then it might incorrectly determine that the channel is idle, and become the second signalling master on the channel.

In peer-to-peer direct operation, collisions between signalling masters can arise for further reasons, such as:

• Movement of MSs: A receiving MS might move within range of a second signalling master, or a signalling master might move within range of another signalling master.

• Hidden node: An MS that wishes to claim the channel might be out of range of another signalling master, but within range of its receiver. This is known as the "hidden node" problem, as the presence of the signalling master is hidden from the listening MS. When the new MS becomes signalling master, the receiver will find itself within range of two signalling masters.

The following table shows which collisions can occur in each of peer-to-peer direct and peer- to-peer repeater operation.

Figure 1 shows a collision occurring between two signalling masters in peer-to-peer direct operation. MSs A and C are both signalling masters, and they are both within range of MS B. Whether MS B hears A, C, or just non-decodable activity, depends on the relative signal strengths. If B hears C's transmissions, then C is said to have "captured" B's receiver.

MS C may or may not be able to hear MS A directly. If C can not hear A, then A is a "hidden node" from C's point of view.

In a group call, an intruding signalling master (MS C in Figure 2) might simultaneously interfere with more than one receiver. In Figure 2, signalling master C is interfering with MSs Bl and B2, that are both trying to hear signalling master A. MSs B3 and B4 are unaware of the collision.

It could be argued that the loss of a number of group members (MSs Bl and B2 in Figure 2) from an unacknowledged group call does not justify notification of the sender (MS A) and release of the channel. On the other hand, it seems inappropriate for a protocol weakness (an inability of the protocol to avoid or resolve collisions between signalling masters) to cause group members to be lost from a call when they are within range. Throughout the remainder of this document, it is assumed that group members will notify their signalling master when they detect a collision. Figure 3 shows a collision occurring between two signalling masters in peer-to-peer repeater operation. MSs A and C are both signalling masters, and they are both within range of the repeater. Whether the repeater hears A, C, or just non-decodable activity, depends on the relative signal strengths. The repeater will transmit signals according to which MS has captured its receiver. In the figure, A is capturing the repeater, so MS B is not able to hear transmissions by MS C while A is transmitting.

The collision problem is symmetrical. Consider the scenarios shown in Figure 1, Figure 2, and Figure 3. In all cases in which MS A might yield to MS C, there is an equivalent case in which C would yield to A. Additionally, in some cases in which A and B are not aware of the presence of C, C or D might become aware of the presence of A, and yield the channel.

The behaviour of a signalling master in response to a collision with another signalling master can be summarised as follows:

1. If a signalling master decodes t dmgCsLimit (typically 2) CS announcements by another signalling master in consecutive frames (in the reserved domain) or TI/RS opportunities (in the payload domain), then it will silently release the channel, making no further announcements.

2. If a signalling master is announcing that the channel is in the reserved domain, and it detects NonDecodableSlotLimit (typically 8) slots with non-decodable activity in them within one frame, then it will silently release the channel.

3. If a signalling master is announcing that the channel is in the payload domain, and it detects non-decodable activity in one or both slots of NonDecodableTiRsLimit (typically 2) Tl and RS opportunities consecutively, then it will silently release the channel.

4. If a signalling master receives a Release Channel Request PDU addressed to itself, it will silently release the channel. After the signalling master has released the channel, its ability to continue with the service (for example, the voice call, or data transfer) that it was providing to its user will depend on the availability of alternative channels. If there are other channels available, the MS can scan for and claim an available channel, and wait for the receiver or receivers in the call to arrive before continuing. If there are no other channels available, the MS should listen to the channel on which the collision occurred to try to establish the new state of the channel. If the new state of the channel can not be explicitly deteπnined by decoding a CS announcement, then the MS can attempt to reclaim the channel. The MS should not attempt to reclaim the channel before it has observed a certain time-out. A suitable time to wait might be the (4 second) analogue coexistence time. If the former signalling master is unable to find another channel, or reclaim the channel on which the collision occurred, the service that it was providing might itself time out. If this occurs, the MS will cease trying to claim a channel.

The response of a receiver in the reserved domain to an intrusion by another signalling master can be summarised as follows. If an MS that is a receiver in the reserved domain:

1. Detects non-decodable activity in NonDecodableCsLimit (typically 2) consecutive slots that should have had CS announcements by its own signalling master in them, or

2. Decodes mtradmgCsLimit (typically 2) CS announcements, in consecutive frames, by a signalling master other than its current signalling master, or

3. Detects more than NoiiDecodableSlotLimit (typically 8) slots with non-decodable activity in them within one frame,

then it will send a Release Channel Request PDU to its current signalling master to inform it that the channel might be no longer usable.

The response of a receiver in the payload domain to an intrusion by another signalling master can be summarised as follows. If an MS that is a receiver in the payload domain:

1. Detects non-decodable activity in NomJecodableCsLimit (typically 2) consecutive slots that should have had CS announcements by its own signalling master in them, or

2. Detects more than NonDecodableSlotLimit (typically 8) slots with non-decodable activity in them within one frame, or

3. Detects non-decodable activity in one or both slots of NonDecodableTiRsLimit (typically 2) Tl and RS opportunities consecutively, then it will send NonDecodableTiRsLimit (typically 2) Release Channel Request PDUs to its current signalling master to inform it that the channel might be no longer usable. A receiver that detects a collision (or an otherwise unusable channel) notifies its current signalling master by sending the Release Channel Request PDU. The Release Channel Request PDU is addressed to the MS that is or was the receiver's current signalling master. It is not acknowledged, as: 1. The receiver is being collided with, so there is a high chance that the acknowledgement won't be received,

2. The signalling master should do a silent release to rninimise the impact of the collision, and

3. In group calls, there could be multiple receivers sending the Release Channel Request PDU, making acknowledgement difficult. In the payload domain, the Release Channel Request PDU is sent in NonDecodableTiRsLimit (typically 2) consecutive Tl and RS opportunities, according to the frame timing of the receiver's own signalling master. In this way, if multiple receivers in a group call are sending Release Channel Request PDUs, and they collide at the repeater or the signalling master, the signalling master will still release the channel.

In the reserved domain, the Release Channel Request PDU will be sent in a random slot selected from among the next 14 random access opportunities, according to the frame timing of the receiver's own signalling master. 14 is the number of random access slots in the reserved domain frame. For instance, if the receiving MS determines that a collision has occurred halfway through a reserved domain frame, it will select a slot in which to send the Release Channel Request from among the next 14 random access slots, where some of those slots will be in the current frame, and the remainder will be in the next. The normal random access rules are not observed after a collision has occurred, as the MS can no longer be sure of the state of the channel. In some cases, the MS will have to transmit its Release Channel Request over the top of an mtruding signalling master. The MS can not be certain that its signalling master will have heard its transmission.

After the receiving MS has transmitted the Release Channel Request or Requests, its ability to continue with the service (for example, the voice call, or data transfer) that it was providing to its user will depend on the availability of alternative channels. If there are other channels available, the MS can scan them, so that it can re-join the service if its signalling master succeeds in claiming one of the other channels. If there are no other channels available, the MS should listen to the channel on which the collision occurred to try to establish the new state of the channel. If the signalling master succeeds in (eventually) reclaiming the channel, the service can be re-joined.

If the former signalling master is unable to find another channel, or reclaim the channel on which the collision occurred, the service that it was providing might itself time out.

An MS that is only observing a connection (but is not part of it) will never indicate a colUsion to the signalling master by means of a Release Channel Request PDU, as the receiver (or receivers) in the connection may not be affected by the collision that is perceived by the observer. When an observing MS detects a collision occurring on the channel, it will simply consider the channel to be unusable, and continue listening to try to re-establish the correct state of the channel.

The observing MS will obey the rules described below.

If an MS that is observing a connection in the reserved domain:

1. Detects non-decodable activity in NonDecodableCsLimit (typically 2) consecutive slots that should have had CS announcements by the channel's signalling master in them, or

2. Decodes IntmdingCsLimit (typically 2) CS announcements, in consecutive frames, by a signalling master other than the channel's current signalling master, or

3. Detects more than NonDecodableSlotLύnit (typically 8) slots with non-decodable activity in them within one frame,

then it will consider the channel to be no longer usable.

If an MS that is observing a connection in the reserved domain:

2. Detects more than NonDecodableSlotLimit (typically 8) slots with non-decodable activity in them within one frame,

then it will consider the channel to be no longer usable. Figure 4 shows the behaviour of a peer-to-peer signalling master in the reserved domain. An MS becomes a peer-to-peer signalling master in the reserved domain by transmitting an FS/CS by random access. This process is not shown in the state model.

During every frame, the MS progresses through the states Sending FS, Sending CS, and Receiving RA slots. In slots 1 and 2, the MS will transmit the FS and CS announcements, and in slots 4 through 17 it will listen for random access signalling. The MS does not listen for RA signalling in slots 3 and 18.

A collision with another signalling master is detected if a CS announcement is decoded in tradmgCsLimit consecutive frames (e.g. 2 consecutive frames). At the start of each frame (in the state Receiving RA slots), the flag htratogCsInThisFrame is set to false. If, at the end of the frame, the flag is still set to false, then the MS knows that no there has been no intruding CS announcement received. The MS can then reset the IntrudingCsCount to zero.

The result of each random access slot is processed in the state Processing Received j-nformation. If another MS has sent a payload header addressed to the members of this call, then this MS will no longer be the signalling master, but will become a receiver in the call. If a CS announcement is received from an intruding signalling master, then the lrudingCsCount is incremented. If IntrudingCsLimit intruding CS announcements have been received (in consecutive frames), then the channel is silently yielded. The MS becomes an unsynchronised MS, in that it no longer knows the current state of the channel. If a Release Channel Request is received from a receiver in the call, or more than

NonDecodableSlotLimit slots of non-decodable activity are heard on the channel, then the channel is silently yielded, and the MS becomes an unsynchronised MS, so that it will listen to re-establish the state of the channel.

Figure 5 shows the behaviour of a peer-to-peer signalling master in the payload domain. An MS becomes a peer-to-peer signalling master in the payload domain by transmitting a payload header by random access, or by transmitting a payload CS announcement in the idle domain. This process is not shown in the state model.

The superframe structure for the peer-to-peer payload domain is such that slots 1 and 2 are alternately used for an FS/CS transmission by the signalling master, and Tl or RS opportunities in which the signalling master listens to the channel. In the state model, no distinction is made between the Tl and RS opportunities. The MS listens to the channel during both slots of the Tl or RS opportunity, and processes the result. Note that the result can simply be an RSSI measurement, if no signalling is successfully received.

The state Processing TI/RS contains the collision avoidance mechanisms for the signalling master. If data is successfully decoded, or a low RSSI is observed, then there is no collision occurring, and the NonDecodableTiRsCount is reset to zero. Otherwise, the count is incremented, and if it has reached the NonDecodableTiRsLimit, the channel is silently released. At this point, the MS no longer knows the state of the channel, so it will be unsynchronised (not synchronised to any signalling master).

If the signalling master decodes an mtruding signalling master's CS announcements in IritradmgCsLimit consecutive Tl and RS opportunities, it will silently release the channel.

Each time a CS announcement by another signalling master is decoded, the trudingCsCount is incremented. If any other signalling is decoded, or if no signalling is decoded at all, then an intruding CS announcement has not have been received, so the mtrudingCsCount is reset to zero.

When the signalling master has decoded tradmgCsLimit CS announcements by an intruding signalling master in consecutive Tl and RS opportunities, it releases the channel and adopts the frame synchronisation of the intruding signalling master.

If the signalling master has received a payload header, then there must be another signalling master on the channel. The signalling master releases the channel and listens to try to establish synchronisation with the new signalling master.

When the signalling master decodes a Release Channel Request addressed to itself, it releases the channel and listens to try to establish the new state of the channel.

Figure 6 shows the behaviour of a peer-to-peer receiver in the reserved domain. A receiver in the reserved domain decodes the random access slots, and listens to its signalling master's FS/CS announcements in slots 1 and 2 of each frame.

If a CS announcement is received while Processing random access data (in slots 3 to 18 of the reserved domain frame), then it must have been transmitted by an intruding signalling master. The mtrudingCsCount is incremented, and if a number ( tmdingCsLimit) of intruding CS announcements have been received in consecutive frames, a collision is indicated to the receiver's own signalling master by sending a Release Channel Request. The sending of the Release Channel Request is described in the state model of a peer to peer receiver indicating a collision to its signalling master (Figure 8). The mtradingCsInThisFrame flag is used to determine whether intruding CS announcements have occurred in consecutive frames. If no mt ding CS is detected in a given frame, then at the end of the frame the IntrudingCsInThisFrame flag will still be false, and the Intruding CS count is reset to zero (in the state Processing CS).

The receiver also counts the number of slots with a high RSSI that were not decodable. If this number reaches the NoriDecodableSlotLimit, the receiver considers a collision to have occurred, and sends a Release Channel Request to its signalling master.

When Processing CS announcements by the receiver's own signalling master, a number of outcomes are possible. The desired outcome is that a CS addressed to this MS or group is successfully decoded. If any other signalling is decoded, or if there is a high RSSI measured on the channel but no signalling can be decoded, then the NonDecodableCsCount is incremented. After NonDecodableCsLimit consecutive failures to decode the expected CS, the receiver sends a Release Channel Request to its signalling master.

If no channel activity is detected at all where the CS announcement was expected, then the receiver might have moved out of range of its signalling master. After SilentCsLimit consecutive silent CS opportunities, the signalling master is considered to have been lost, and the channel state is no longer known. Figure 7 shows the behaviour of a peer-to-peer receiver in the payload domain. A receiver in the payload domain decodes the payload slots, and listens to the FS/CS announcements and TI RS opportunities in slots 1 and 2 of alternate frames.

When Processing payload slots, the receiver will count the number of slots with a high RSSI that were not decodable. If this number exceeds the NonDecodableSlotLimit, the receiver considers a collision to have occurred, and indicates the collision to its signalling master by sending Release Channel Requests. The sending of these Release Channel Requests is described in the state model of a "Peer to peer receiver indicating collision to signalling master".

When Processing CS announcements by the receiver's own signalling master, a number of outcomes are possible. The desired outcome is that a CS addressed to this MS or group is successfully decoded. If any other signalling is decoded, or if there is a high RSSI measured on the channel but no signalling can be decoded, then the NonDecodableCsCount is incremented. After NonDecodableCsLimit consecutive failures to decode the expected CS, the receiver indicates to its signalling master that a collision might have occurred, by sending Release Channel Requests.

If no channel activity is detected at all where the CS announcement was expected, then the receiver might have moved out of range of its signalling master. After SilentCsLimit consecutive silent CS opportunities, the signalling master is considered to have been lost, and the channel state is no longer known.

An mt ding signalling master is detected while Processing TI/RS opportunities if the receiver detects high RSSI (caused by payload transmissions, or by detecting part of an intruding CS announcement), or decodes an intruding CS announcement directly. In either of these cases, the receiver will cease decoding payload until it has indicated the collision to its signalling master.

Figure 8 shows the behaviour of a peer-to-peer receiver indicating a collision to its signalling master. When a peer-to-peer receiver determines that there has been a collision, it ceases listening to the channel until it has sent its Release Channel Request or Requests. This is particularly important in the payload domain, as the receiver can not be sure who the intended recipient is of any payload that it receives.

When a collision has occurred in the payload domain, the MS calculates when the next Tl or RS opportunity of its signalling master will be, and transmits the Release Channel Request in that opportunity. This process is repeated until the MS has sent the correct number (NonDecodableTiRsLimit) of Release Channel Requests. The MS is then no longer synchronised with whatever channel activity remains, and will listen to establish the new state of the channel.

In the reserved domain, the MS chooses a random slot between slot 4 and slot 17 (inclusive) in which to send the Release Channel Request. If the chosen slot number has already passed in this frame, then the Release Channel Request will be sent in the chosen slot of the next frame. In the reserved domain, the Release Channel Request is sent only once, and the MS will then listen to establish the new state of the channel.

Figure 9 shows a signalling master, A, in the reserved domain, hearing another signalling master, C, that is also in the reserved domain. A collision such as this can occur due to movement of MSs, or due to a false assessment of the channel domain by one of the MSs. The collision is resolved in the following way:

[1] A is transmitting FS/CS announcements.

[2] During A's RA slots, it decodes a number of CS announcements by C in consecutive frames. This number is defined as the tmdmgCsLirnit, and might typically be 2.

[3] A silently yields the channel. A will not make an idle announcement, as the channel is not idle.

MS C in Figure 10 is within range of receivers Bl and B2, that are listening to signalling master A. This situation can arise due to movement of the MSs, or because MS A is "hidden" from MS C. The collision is resolved in the following way:

[1] A is transmitting FS/CS announcements.

[2] During the random access opportunities, a number of CS announcements by C (IntradmgCsLimit) are heard by MSs Bl and B2.

[3] Bl and B2 notify A of the collision in a random access opportunity, by sending a Release Channel Request.

[4] A silently yields the channel.

When the receivers in the reserved domain attempt to send a Release Channel Request to their signalling master, there is a small chance that they will collide. After sending the Release Channel Request, they will listen to the channel to determine its new status. If their Release Channel Requests collided, they will decide that there is (still) a collision occurring, and will once again attempt to resolve it.

An MS can become the new speaker in a call during the reserved domain. If the current signalling master is announcing that the channel is in the reserved domain, then the new speaker transmits a payload header by random access, and assumes that it is now the new signalling master. If the old signalling master fails to decode the payload header, then both MSs will believe themselves to be the current signalling master. Figure 11 shows the case in which MS C becomes a new speaker in a call, but MS A, the former signalling master, fails to decode the payload header. The collision is resolved in the following way: [1] A is announcing that the channel is in the reserved domain.

[2] C transmits a payload header by random access, but A does not decode it. C fills the remainder of the frame with payload slots. A is not able to decode the payload slots, as it is expecting random access transmissions, which use a different channel coding from payload.

[3] Both A and C make CS announcements, so they do not hear each other.

[4] A detects more than a certain number of slots within one frame that have non- decodable activity in them. This number is the NonDecodableSlotLimit, and might typically be 8. [5] A silently yields the channel. If A was to make an idle announcement here, then

MSs receiving the call would incorrectly conclude that the call had ended.

[6] A hears C's CS announcement, and determines that it should now be a receiver in the call.

Figure 12 shows a scenario in which a signalling master, C, in the payload domain, intrudes upon a receiver, B. B is hstening to signalling master A, that is in the reserved domain. This collision could occur due to movement of MSs, false assessment of the channel domain, or a hidden node problem. The collision is resolved in the following way:

[1] The CS announcement transmitted by A may or may not be received by B, depending on the relative levels of the signals from A and C.

[2] B decodes a CS announcement by C. One CS announcement by an mtrading signalling master is not enough for B to conclude that a collision is occurring.

[3] By the end of the frame (according to A's frame timing), B has received non- decodable activity in more than a certain number of slots (NoriDecodableSlotLimit), so it concludes that a collision has occurred. [4] B transmits a Release Channel Request PDU by random access in the next frame, according to the frame timing of its own signalling master, A.

[5] A silently yields the channel. Even if the receiver, B, can no longer hear its own signalling master, A, it still knows the timing of A's frames. B will transmit its Release Channel Request according to its own understanding of A's frame timing. B can not strictly observe the random access rules to send the Release Channel Request, as these rules would dictate that B should not transmit at all, due to the channel activity that is being transmitted by C. When a receiver needs to transmit a Release Channel Request in the reserved domain due to a collision, it will do so in a random slot selected from among the next 14 random access opportunities, according to the frame timing of its own signalling master.

Figure 13 shows a signalling master, C, in the payload domain, intruding upon a signalling master A, that is also in the payload domain. The collision might have occurred due to movement of the MSs, or a false assessment of the channel domain, and the two signalling masters are not synchronised with each other. An alternative case, in which the signalling masters are synchronised, is considered in Figure 15. The collision in Figure 13 is resolved in the following way:

[1] A is transmitting its own FS/CS announcement and payload slots, so it does not hear

C.

[2] During A's Tl opportunity, it detects a high RSSI level. A fails to decode anything, as it is not expecting payload slots. Generally, one burst of non-decodable activity is not enough for A to release the channel, so it continues as signalling master. [3] In A's next RS opportunity, it once again detects a high RSSI level, and fails to decode anything. A now decides that the channel is possibly unusable, as it has heard non-decodable activity in a certain number of consecutive Tl or RS opportunities. This number is the NonDecodableTiRsLimit, and might typically be 2.

[4] A silently yields the channel.

Figure 14 shows a signalling master, C, in the payload domain, mtmding upon a receiver B, that is listening to payload from signalling master A. The two signalling masters are not synchronised. This collision might occur due to movement of the MSs, false assessment of the channel domain, or a hidden node problem. The collision is resolved in the following way: [1] B becomes aware of the intrusion due to a failure to receive a certain number of consecutive CS announcements by A. This number is the NonDecodableCsLimit, and might typically be 2.

[2] B transmits a Release Channel Request, addressed to A, in A's next Tl or RS opportunity. B has no way of knowing whether A successfully decoded this telegram.

[3] B transmits a further Release Channel Request, addressed to A, in A's next Tl or RS opportunity. An affected receiver will transmit Release Channel Requests in a number (NonDecodableTiRsLiinit) of consecutive Tl and RS opportunities.

[4] A either decodes the Release Channel telegram, or measures a high RSSI in a number of consecutive Tl or RS opportunities (Noiύ ecodableTiRsLimit consecutive opportunities), so it silently yields the channel.

If B (or the repeater) is not captured by C in Figure 14, then B will only be aware of C due to non-decodable activity in the Tl and RS opportunities. Although there is a collision happening, B is not affected, so no action need be taken.

Figure 15 shows a collision caused by simultaneous transmission of payload headers in a group call. A resolution of the collision is not assured, as the colliding parties will not hear each other, and other parties in the call may not be able to decode either of the colliding parties either. During individual calls, the collision will only be resolved when one of the parties voluntarily ceases transmission. Figure 15 shows that in group calls, the collision will commonly be resolved fairly promptly, as follows:

[1] Signalling master A is announcing that the channel is in the reserved domain.

[2] MSs B and C simultaneously, or nearly simultaneously, transmit payload headers. These payload headers collide at the repeater, so no other MS is able to decode them or any of the following payload.

[3] Neither A, B, nor C are aware that the collision has occurred, so they all transmit their CS announcements.

[4] After receiving a certain number of slots (NonDecodableSlotLimit) of non- decodable channel activity, A becomes aware of the collision. [5] A silently yields the channel.

At this point, neither A nor any other member of the group is aware that the colUsion between B and C is occurring. All they know is that there is non-decodable activity on the channel. No MS other than B or C is aware of the frame timing of B or C, so no MS is able to transmit a Release Channel Request in the Tl or RS opportunities of B and C. The collision will eventually be resolved however, as follows:

[6] A variation in the propagation conditions (due to shadowing or fading, for instance) allows the CS announcement by B to be received by the repeater, and consequently by MS A. MS A now believes that the current signaUing master is B, and expects to receive payload from B.

[7] A receives non-decodable activity in more than a certain number of slots (NonDecodableSlotLimit) during the next frame.

[8] A sends a Release Channel Request in the next Tl or RS opportunity, addressed to B. B silently yields the channel.

In the above scenario, MS A will send NonDecodableTiRsLimit Release Channel Requests in consecutive Tl and RS opportunities. This has not been shown in Figure 15 for the sake of simplicity.

Commonly in group calls, multiple MSs will send Release Channel Requests in consecutive Tl and RS opportunities. When this occurs, it is likely that both of the colliding MSs will release the channel. They will then be able to compete again to become signalling master, and normal random access procedures should prevent them colliding again.

When a signalling master, such as MS A in Figure 16, is in the payload domain, it only listens to the channel during its own Tl and RS opportunities. When there is a collision with another signaUing master (MS C, for instance) that is in the reserved domain, MS A will only be directly aware of the collision if it can hear MS C's announcements in its Tl and RS opportunities. Figure 16 illustrates a number of different timing possibilities for such a collision. The top line shows MS A's transmission of payload data, followed by a 2 slot Tl or RS opportunity. The remaining lines show a number of possibilities when MS C's FS/CS bursts land on or near the Tl or RS opportunities of MS A. In the figure, the CTS parts of these FS/CS bursts are not labelled. The cases illustrated are: [1] MS C's FS/CS burst does not overlap with the Tl or RS at all, so MS A is completely unaware of the presence of MS C.

[2] MS C's FS/CS burst falls too early for MS A to receive the MSS part of the burst. MS A can not synchronise with MS C, so it is unable to decode the CS signalling. MS A measures a high RSSI level on the channel in slot 1 of the Tl or RS opportunity. When this happens during a number of consecutive Tl and RS opportunities (NonDecodableTiRsLimit), A will yield the channel.

[3] MS A is able to synchronise to the MSS transmitted by MS C, but loses the end of the CS telegram. MS A measures a high RSSI level on the channel in slot 2, and potentially also in slot 1. When this happens during a number of consecutive Tl and RS opportunities (NonDecodableTiRsLimit), A will yield the channel.

[4] The FS/CS burst falls too late for MS A to gain synchronisation at all. MS A measures a high RSSI level on the channel in slot 2. When this happens during a number of consecutive Tl and RS opportunities (NonDecodableTiRsLimit), A will yield the channel.

[5] Even though the entire FS/CS burst does not fit within the Tl or RS opportunity, MS A is able to synchronise to the transmission and decode the CS telegram. MS A learns that it has collided with another signalling master when this occurs in a number (IntrudingCsLimit) of consecutive Tl and RS opportunities, and will yield the channel.

The timings of the FS/CS bursts illustrated in Figure 16, and the cases described above, apply equally to a receiver that hears the FS/CS bursts of the intruding signalling master. When the receiver learns that a collision has occurred, due to repeated non-decodable activity or intruding CS announcements in the Tl and RS opportunities, or due to loss of its own signalling master's CS announcements, it will attempt to notify its own signalling master of the collision by sending a Release Channel Request in a number (NonDecodableTiRsLimit) of consecutive Tl and RS opportunities.

Figure 17 shows a typical mobile unit used in DIIS or other radio communications, in which some of the power related components have been indicated. The unit includes a main controller 10 that is responsible for most functions of the unit, and a transmit power controller 11 that is generally part of the main controller. Program instructions and data required by the controller are stored in memory 12. A rechargeable battery 13 powers the unit. The main controller is connected by a bus 14 to the battery, a keypad 15, display 16 and a battery charge indicator 17.

The radio unit in Figure 17 includes transmitter and receiver blocks 20, 21 connected to an antenna through a coupler 23. Either simplex or duplex operation may be possible. A power amplifier 24 determines the output magnitude of the transmitter, both under control of the power controller 11. Specific power saving schemes may be implemented by the power controller and the main controller, by varying the transmitter output magnitude and power on/off state of the radio as a whole. An audio processing component 26 connects a microphone and a speaker to the transmitter and receiver respectively.

It will be appreciated that power saving systems according to the invention are not limited to DIIS and that the terminology, systems and principles described here in relation to DIIS have relevance to digital radio systems in general.

Claims

CLAIMS:

1. A method of operating a mobile unit in a peer to peer radio communication system comprising: announcing control of a channel in the system, transmitting on the channel to another unit, detecting a possible collision on the channel, and releasing the channel.

2. A method according to claim 1 wherein the collision is detected as an announcement on the channel by another unit or as non- decodable activity on the channel.

3. A method of operating a mobile unit in a peer to peer radio communication system comprising: receiving an announcement of control by another unit of a channel in the system, receiving a transmission from the other unit on the channel in accord with the control, detecting a possible collision on the channel, and transmitting a warning of the collision to the other unit.

4. A method according to claim 2 wherein the collision is detected as an announcement on the channel by another unit or as non- decodable activity on the channel.

5. A mobile unit for a radio communication system that is capable of operating a method according to any one of claims 1 to 4.

6. A radio communication system having mobile units that operate a method according to any one of claims 1 to 4.