WO2015055243A1

WO2015055243A1 - Scheduling shared spectral resource in mobile communications system

Info

Publication number: WO2015055243A1
Application number: PCT/EP2013/071653
Authority: WO
Inventors: Patrick SHEEHY; Ying Sun
Original assignee: Telefonaktiebolaget L M Ericsson (Publ)
Priority date: 2013-10-16
Filing date: 2013-10-16
Publication date: 2015-04-23

Abstract

A method of scheduling resources, such as LTE resource blocks, or tones, for a cell of a cellular radio communication system serving multiple mobile user equipments UEs (120), that involves scheduling a shared spectral resource to share it between the mobile UEs, and when detecting (210) at a MAC layer, such as over a Channel Quality Indication, CQI, or a data error rate, for a mobile UE that its radio channel condition has reached a threshold of deterioration indicating failure of transmission, re-scheduling its resources such that its share of the spared spectrum resource can be used more efficiently by others. If so, its share of the shared spectral resource is reduced (220) without waiting for release from the cell of the said at least one of the mobile UEs, and the shared resource no longer scheduled is made available for others of the mobile UEs. This increases the efficiency and overall throughput over the shared spectral resource.

Description

SCHEDULING SHARED SPECTRAL RESOURCE IN MOBILE

COMMUNICATIONS SYSTEM Field

The present invention relates to methods of scheduling, and to base station apparatus for scheduling, and to corresponding computer programs.

Background

It is known to provide cellular wireless communication networks with a scheduler to share out a shared spectrum resource between different UEs in a given cell. One example is LTE networks which have an IP-based network architecture and OFDM-based radio channels. The base station in the form of an eNodeB (eNB) has a protocol stack which includes a MAC sublayer. This has a MAC Scheduler which schedules the downlink/uplink transmissions and allocates physical layer resources to be used. A link adaptation module is used to adapt the coding and data rate to the radio channel conditions, so that as the link quality degrades, a reduced amount of data is sent in each scheduled transmission slot.

According to US 2012294258 a scheduler can maximise utility in scheduling involving granting spectral resources to mobile terminals (typically in every slot), and it may select the rate and power at which the mobile terminals may transmit. The allocation decisions may maximize a sum of utilities where utility is based on respective QoS Class Identifier (QCI) parameters of the different logical channels, or respective prioritized bit rate (PBR), or the priority of the logical channel, and is subject to constraints given by, for example, the inter-cell interference management block 408. Therefore, information available to the scheduler may comprise at least one of: the interference constraints, the buffer state, or the power headroom.

It is known that radio channel conditions in a cellular communication network can vary due to inter cell interference or other interference. Depending on the quality of coverage of a radio network, some communication between a user equipment UE and a base station apparatus such as an eNB, can fail. For example, a UE can enter a black spot which can vary in size and/or time, where the radio signal is too weak to transport data successfully. Such a UE can either regain coverage after a short period of time as it moves out of the black spot, or, if the problem persists, drop this connection to the eNB and try to reconnect later on.

Another cause of poor radio channel conditions is disturbance or interference from another source that is so large that the signal from the UE cannot be detected in the eNB. High inter cell interference may also result in the eNB not detecting a UE's transmission.

For such periods when poor channel conditions mean that a UE temporarily does not have a radio connection capable of transporting data, the network reacts to manage this. A number of functions already exist in an eNB to react to such scenarios. The main ones are as follows:

1 . Inactivity timer

The eNB supervises activity on each data radio bearer. If no data has been sent or received for a configurable period of time, a decision is taken to release the UE.

2. Out of Sync

The eNB supervises activity on a UE and if no data has been sent or received for a configured period of time, that is longer than out-of-sync timer, but shorter that inactivity timer, it is decided that the UE is out-of-Sync.

3. Radio Link Failure

All DL transmissions for RLC Acknowledgment Mode bearers must receive an ACK PDU on UL. ACK PDU's which are not received within a configured period of time trigger a release of the UE.

Summary

Embodiments of the invention provide improved methods and apparatus. According to a first aspect of the invention, there is provided a method of scheduling for a cell of a cellular radio communication system serving mobile user equipments UEs, by scheduling a shared spectral resource to share it between the mobile UEs, and detecting at a media access control protocol layer for at least one of the mobile UEs whether an indication of its radio channel condition has reached a threshold of deterioration indicating failure of transmission over a period of time. If so, a reduced share of the shared spectral resource is scheduled for that mobile UE without waiting for release of that mobile UE from the cell. At least some of the shared spectral resource no longer scheduled for the said at least one mobile UE, is made available for others of the mobile UEs.

A benefit of the reduced share is reduced wastage of spectral resource or processing resource or UE battery power while waiting for other layers of protocol stack to decide whether to release the UE, or while the UE is declared out of sync, for example, which can take hundreds of msecs. By detecting at the MAC layer, the threshold of deterioration can be determined more quickly than the time taken to release the UE, which usually would be controlled at a higher protocol layer, and thus take longer. This detection is intended to encompass detecting at lower protocol layers, which detections will be passed up to the MAC layer and thus detected at that layer. In some cases, the spectrum resource being wasted can be used more efficiently by other UEs. Even if there is no reuse of the shared spectrum resource, there can still be a reduction in transmitter power consumption and reduced processing. Such power consumption reduction can lead to longer battery life in the UE in particular. The reduced share is compatible with existing link adaptation techniques for altering coding to reduce data transmission rate as radio channel conditions deteriorate. This threshold of deterioration can be a fixed or dynamic level, in principle, and could be dependent on performance of other radio channels, or on a relative priority level, relative to the others, so that the shared spectrum resource is used more efficiently. The threshold and period of time can be set according to to how much of the wastage is to be saved. See fig 2 for example.

Any additional features can be added to any of the aspects, and some such additional features are set out below. One such additional feature is the step of scheduling shared spectral resource comprising scheduling time slots and frequencies, and the reduced share comprising at least one of: fewer of the time slots, a reduced size of the time slots and fewer of the frequencies. This is a convenient way to divide spectral resources, suitable for various widely used standards. Other ways are conceivable such as code division, and combinations of these and other ways of dividing the spectrum. See fig 3 for example.

Another such additional feature is the step of scheduling with a reduced share comprising scheduling sufficient of the shared spectral resource for detection and reporting of the radio channel condition for the said at least one of the mobile UEs. This smaller share can help reduce the wastage problem of larger shares scheduled for failing radio channels, while still enabling rapid increase in share when the deteriorated channel conditions improve as appropriate, to reduce needless latency. See fig 9 for example.

Another such additional feature is the step of scheduling shared spectral resource comprising scheduling according to an indication of level of data in buffers awaiting transmission, and the step of scheduling a reduced share comprising: reducing the indicated level of data in buffers in respect of the said at least one UE to cause the reduced share, or responding to a new request for transmission by increasing the indication of level of data by less than a typical increment, or reducing a priority of the data in buffers relative to that of the others of the UEs, or reducing a dependency of the share for the said at least one UE on its respective indicated level of data in buffers. These are convenient ways of influencing the scheduling and making more spectral resource available to others. See fig 4 for example.

Another such additional feature is the detecting when the threshold has been reached comprising detecting whether the cell is overloaded and adapting the threshold if the cell is overloaded, to cause a worst performing one of the radio channels to reach the threshold of deterioration. If the cell is overloaded (e.g not meeting QoS for all requests) then there is additional benefit in reducing the share of the least efficient radio channel, since the shared spectral resource freed up will be used by other UEs. Hence it is worth adapting the threshold to increase the likelihood of the reduced share.

Another such additional feature is the step of scheduling shared spectral resource comprising scheduling according to: a predetermined sequence, or a predicted demand, and the step of scheduling the reduced share comprising: reducing the share scheduled according to the predetermined sequence, or adapting the sequence to cause the reduced share, or reducing the share scheduled according to the predicted demand, or adapting the prediction of demand to cause the reduced share. This can help ensure the reduction applies also to such scheduling as well as on demand scheduling. Such sequences can help reduce scheduling overhead by avoiding or reducing the need for individual scheduling requests. By reducing delays caused by such overhead, latency can also be reduced. Hence applying the reduction to such scheduling is useful to reduce possible wastage further. Scheduling according to predictive demand can reduce latency caused by sending requests from the UE and, calculating and sending out schedules in response. There can be external sources of predicted demand such as service aware functions for example See fig 5 and fig 12 for example.

Another such additional feature is a step of operating in a poor channel state according to the indication of the radio channel condition being worse than the threshold, and having the step of exiting the poor channel state according to an indication of the radio channel being better than a second threshold, with hysteresis provided between the entry into and the exiting from the poor channel state. Having such a state and having hysteresis between entry and exit can help to avoid instability such as toggling between states. The hysteresis can encompass using different thresholds, or providing time delays for example. See fig 6 and fig 7 for example.

Another such additional feature is the detection step being based on at least one of: detected energy received over the radio channel, and a result of a CRC step. Either or both of these can help enable a rapid and reliable detection. A benefit of using both is that they are different types of indications and so more information can be obtained. See figs 7 and 8 for example.

Another such additional feature is the scheduling comprising centralized medium access control scheduling of logical channels of an LTE system and the step of reducing comprising reducing the scheduling of at least one of: assignments of time slots and frequencies for downlink logical channels onto a downlink shared transport channel and grants of time slots and frequencies for uplink logical channels onto an uplink shared transport channel. Such LTE MAC scheduling is commercially a particularly valuable application, though the benefits can be apparent in applications to other systems. See fig 10 for example.

Another aspect provides acomputer program on a non transitory computer readable medium and having instructions which when executed by a computer, cause the computer to carry out the methods set out above.

Another aspect provides base station apparatus for a cell of a cellular radio communication system for serving mobile user equipments UEs, the base station apparatus having processing means operative to schedule a shared spectral resource to share it between the mobile UEs, and to detect at a media access control protocol layer, for at least one of the mobile UEs, whether an indication of its radio channel condition has reached a threshold of deterioration indicating failure of transmission over a period of time. If so, the processor is operative to schedule a reduced share of the shared spectral resource for that one of the mobile UEs without waiting for release from the cell of the corresponding mobile UE, and to make at least some of the shared spectral resource no longer scheduled for that mobile UE, available for others of the mobile UEs. This corresponds to the method above and has corresponding benefits. See figs 1 and 2 for example.

Another such additional feature is the processing means comprising a processor and a memory, said memory containing instructions executable by said processor to carry out the scheduling. See fig 1 for example.

Another such additional feature is the processing means being operative such that the scheduling of the shared spectral resource comprises scheduling time slots and frequencies, and the reduced share comprises at least one of: fewer of the time slots, a reduced size of the time slots and fewer of the frequencies. Again this corresponds to the above methods, and has corresponding benefits. See figs 1 and 3 for example.

Another such additional feature is the processing means being operative to carry out the scheduling with a reduced share while scheduling sufficient of the shared spectral resource for detection and reporting of the radio channel condition for the said at least one of the mobile UEs. Again this corresponds to the above methods, and has corresponding benefits.

Another such additional feature is the processing means being operative such that the scheduling of shared spectral resource comprises scheduling according to an indication of level of data in buffers awaiting transmission, and the step of scheduling a reduced share comprises at least one of: reducing the indicated level of data in buffers in respect of that UE to cause the reduced share, responding to a new request for transmission by increasing the indication of level of data by less than a typical increment, reducing a priority of the data in buffers relative to that of the others of the UEs, and reducing a dependency of the share for that UE on its respective indicated level of data in buffers. Again this corresponds to the above methods, and has corresponding benefits.

Another such additional feature is the processing means being operative to detect when the threshold of deterioration has been reached by detecting whether the cell is overloaded and being operative to adapt the threshold if the cell is overloaded, to cause a worst performing one of the radio channels to reach the threshold of deterioration. Again this corresponds to the above methods, and has corresponding benefits.

Another such additional feature is the processor being operative to schedule the shared spectral resource by scheduling according to at least one of: a predetermined sequence, and a predicted demand, and the processor also being operative to schedule the reduced share by at least one of: reducing the share scheduled according to the predetermined sequence, adapting the sequence, reducing the share scheduled according to the predicted demand, and adapting the prediction of demand. Again this corresponds to the above methods, and has corresponding benefits.

Another such additional feature is the processing means being operative in a poor channel state according to the indication of the radio channel condition being worse than the threshold, and being operative to exit the poor channel state according to an indication of the radio channel condition being better than a second threshold, with hysteresis provided between the entry into and the exiting from the poor channel state. Again this corresponds to the above methods, and has corresponding benefits.

Any of the additional features can be combined together and combined with any of the aspects. Other effects and consequences will be apparent to those skilled in the art, especially compared to other prior art. Numerous variations and modifications can be made without departing from the claims of the present invention. Therefore, it should be clearly understood that the form of the present invention is illustrative only and is not intended to limit the scope of the present invention.

Brief Description of the Drawings:

How the present invention may be put into effect will now be described by way of example with reference to the appended drawings, in which:

Fig 1 shows a schematic view of apparatus according to an embodiment, Fig 2 shows operational steps according to an embodiment,

Fig 3 shows operational steps according to another embodiment with scheduling in terms of frequencies and time slots,

Fig 4 shows operational steps according to another embodiment with scheduling according to buffer levels,

Fig 5 shows operational steps according to another embodiment with the reduced scheduling applied also to scheduling made according to predetermined sequences and/or predicted demand,

Fig 6 shows operational steps according to another embodiment with entry to and exit from a poor channel state,

Fig 7 shows a state diagram showing a poor channel state,

Fig 8 shows operational steps according to another embodiment with a threshold for reduced scheduling based on DTX and/or CRC

Fig 9 shows operational steps according to another embodiment with the reduced scheduling still allowing detection and reporting of radio channel conditions,

Fig 10 shows a schematic view of a protocol stack for a base station apparatus including an expansion of a MAC layer, showing a scheduler, according to an embodiment,

Fig 1 1 shows operational steps according to another embodiment with MAC layer scheduling of logical channels of an LTE system,

Fig 12 shows a state diagram showing a talk state and a silent state for a VoIP call example, according to embodiments, and

Fig 13 shows steps according to another embodiment with reduced share prompted by cell overload. Detailed Description:

The present invention will be described with respect to particular embodiments and with reference to certain drawings but the invention is not limited thereto but only by the claims. The drawings described are only schematic and are non- limiting. In the drawings, the size of some of the elements may be exaggerated and not drawn to scale for illustrative purposes.

Where the term "comprising" is used in the present description and claims, it does not exclude other elements or steps and should not be interpreted as being restricted to the means listed thereafter. Where an indefinite or definite article is used when referring to a singular noun e.g. "a" or "an", "the", this includes a plural of that noun unless something else is specifically stated.

Elements or parts of the described apparatus, nodes networks may comprise logic encoded in media for performing any kind of information processing. Logic may comprise software encoded in a disk or other computer-readable medium and/or instructions encoded in an application specific integrated circuit (ASIC), field programmable gate array (FPGA), or other processor or hardware.

References to base station apparatus can encompass any kind of base station, for any standard, not limited to any level of integration, or size or bandwidth or bit rate and so on, unless required by the context.

References to programs or software can encompass any type of programs in any language executable directly or indirectly on processing hardware.

References to processors, hardware, processing hardware or circuitry can encompass any kind of logic or analog circuitry, integrated to any degree, and not limited to general purpose processors, digital signal processors, ASICs, FPGAs, discrete components or logic and so on. References to a processor are intended to encompass implementations using multiple processors which may be integrated together, or co-located in the same node or distributed at different locations for example.

Reference to sharing of spectral resource is intended to cover at least sharing by time or by frequency or other dimension.

Reference to indication of channel condition is intended to encompass measurements, indirect indications, DTX indication, data error rates, loss of signal indications, combinations of these, statistics of these, and so on. Reference to reduced share is intended to encompass reduced number of frequencies, reduced number or size of time slots allocated, reduced priority level relative to other UEs to cause a reduction in share and so on. A reduced size of time slots is intended to encompass the future possibility that time slots can have different durations, which could enable choosing a reduced duration time slot as a way of reducing the share, even if that is not typical with the current standards.

Reference to levels is intended to encompass at least indications of local buffer levels and estimates of levels of remote buffers.

Reducing is intended to encompass partial reduction or complete reduction to zero if the context allows.

Reference to detected energy is intended to encompass detection either before or after decoding, and either as signal energy or signal over noise, or a ratio of signal and noise and so on.

Abbreviations:

CRC Cyclic Redundancy Check

DL downlink

DTX Discontinuous Transmission

HARQ Hybrid Automatic Repeat Request

PDU Protocol Data Unit

RLC Radio Link Control

SB Scheduling Block SR Scheduling Request

SPS Semi Persistent Scheduling

SE Scheduling Entity (A UE that shall be allocated spectrum

this scheduling opportunity)

UE User Equipment

UL uplink

By way of introduction to the embodiments, how they address some issues with conventional designs will be explained. Until now it has not been appreciated that with existing scheduling, if communication to a UE is failing, the eNB can waste valuable amounts of the shared spectrum resources by continuing to schedule for a deteriorated radio channel while waiting for one of the existing supervision functions to react to the deteriorated radio channel and downgrade or release the corresponding UE.

Amongst the shared spectrum resources which can be wasted (using LTE MAC terminology) are:

SBs on PUSCH (physical uplink shared channel)

SBs on PDSCH (physical downlink shared channel)

CCE's on PDCCH (physical downlink control channel)

Furthermore other resources can be wasted such as Baseband processing power (at UEs or at eNB), and transmitter power, which both contribute to battery power wastage at the UEs.

The typical delays in reacting are as follows: For the inactivity timer a decision is taken to release the UE after a time of typically 10 seconds. For the out of sync decision, a typical time is 1920ms. For Radio Link Failure, a typical time for RLC retransmissions varies from 400 - 640ms (and even higher).

Out-of-Sync is the least intrusive of these functions in that it only requires a new Random Access before user data can be sent/received to this UE. In order to balance latency requirements for transmitting user data and load on PRACH (physical random access channel) the time window for out-of-sync is quite large.

Both Inactivity and Radio Link Failure lead to the release of a UE from the eNB. However releasing a UE that has had a temporary loss of coverage can generate even more load, as it will perform re-establishment or establish a new connection as well as the added latency experienced by the user. For this reason the time window configured on these supervision functions tends to be large, with the drawback that they can continue to consume resources during this time.

If the radio channel condition to a UE is poor and many transmissions fail, the link adaptation will adjust to more conservative settings using more CCE's on PDCCH, as well as reducing the amount of data sent on PUSCH/PDSCH for each transmission. In addition each transmission can be scheduled with a max number of HARQ attempts. All these actions magnify the resource wastage when the UE has poor radio channel conditions.

This wastage can occur in the following situations for example:

1 . The eNB has data in UL buffer when the UE loses coverage which it will continue to grant until either the UE regains coverage or it is released.

2. The eNB has data in DL buffer when the UE loses coverage which it will continue to grant until either the UE regains coverage or it is released.

3. Data is periodically added to a UE's UL buffer by an internal eNB function triggering UL grants e.g., periodic SR, SPS or other enhanced buffer estimation functions that are common in low latency services e.g. VoLTE.

4. Data is periodically added to a UE's DL buffer by an internal eNB function triggering DL transmissions e.g. timeAlignment adjustments.

In the unlucky scenarios this data can be in a buffer which results in these UE's being prioritized over other UEs. In a resource limited eNB this will lead to reduced capacity and unnecessary delays for other UEs data.

Figs 1 ,2, Features of embodiments of the invention

Embodiments of the invention can extend the scheduler to detect poor radio channel conditions, which are causing such wastage and are likely to lead to actions at other layers such as release or out-of-sync as described above. The scheduler can then reduce the amount of shared spectrum resource scheduled for such poor radio channels, without waiting for existing steps such as UE release to occur, before making the share no longer used available for increasing the shares of others of the UEs. This can reduce wastage of the shared spectrum resource and other wastage. Figure 1 shows a schematic view of a base station 100 for a cell of a cellular wireless communications system, configured to communicate over radio channels to UEs 120. The base station has a scheduler 1 10 implemented by a processor and memory 95. The scheduler has detector 130 for detecting that radio channel conditions have reached a threshold of deterioration. The scheduler also has a module 140 for scheduling with a reduced share, triggered by an output of the detector, to reduce the scheduling for any UE or UEs having such poor radio channel conditions, without waiting for existing functions to react.

Thus such wastage of shared spectral resource can be reduced instead of waiting while other layers of protocol stack decide whether to release the UE, or while the UE is declared out of sync, for example, which can take hundreds of msecs. Instead the reduced scheduling can take place in time scales of one or a few time slots, which can be in the order of l Omsecs or less. Thus the threshold can correspond to such a deterioration in transmission that the spectrum is being wasted and could be more efficiently used by other UEs. This threshold can be a fixed or dynamic level, in principle. Sharing of spectral resource is intended to cover at least sharing by time or by frequency or other dimension. An "indication of channel condition" is intended to encompass measurements, indirect indications, DTX indication, data error rates, loss of signal indications, combinations of these, statistics of these, and so on. There are many variations possible in implementing the detector and in implementing the module for reducing scheduling, and some will be described in more detail below.

Figure 2 shows some operational steps of a scheduler according to an embodiment, using the features of claim 1 or of other embodiments. At step 200, there is a normal operating step of scheduling shared spectral resources to share between UEs at least according to criteria such as their priorities. Step 209 shows an optional step of providing link adaptation for one of more UE to adapt its data rate within its scheduled share of the shared spectral resource, according to its radio channel condition. Step 210 shows detecting at the MAC layer for each UE whether an indication of its radio channel condition has reached a threshold of deterioration indicating failure of transmission over a period of time. This indicates inefficient use of the shared spectrum. The period of time can be selected as desired to ensure sufficiently rapid detection, but avoid isolated transient transmission failures. If yes, then at step 220 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE. Reduction can mean a partial reduction or in some cases a complete reduction to stop any scheduling. Step 230 shows making available that shared spectral resource no longer used by that UE, for scheduling for other UEs to increase their share of the shared spectrum resource.

These steps can be implemented in various ways or other steps can be added. In some embodiments the scheduler can be extended with features such as:

1 . a quicker detection of unstable radio connections compared to the existing functions.

2. A new state representing poor channel state (to provide robustness to such conditions) where existing buffer sizes in the scheduler are set to zero and functions that periodically inject data to these buffers are turned off e.g. enhanced buffer estimation functions for VoLTE.

3. A periodic UL granting of the UE can be performed allowing it a quick method to return to normal scheduling if the radio link stabilizes. These grants can be issued with a lower frequency than would have been used in the 'normal grant behavior' and the grant size can be minimal e.g. sufficient for detecting and reporting improving radio channel conditions, such as by using existing protocols (BSR) for reporting levels of data in buffers awaiting transmission from the UE.

Fig 3, embodiment with scheduling in terms of frequencies and time slots

Figure 3 shows a variation of the embodiment of figure 2 and shows operational steps of a scheduler using the same reference numbers as figure 2 where appropriate. At step 201 , there is a normal operating step of scheduling shared spectral resources to share between UEs at least according to criteria such as their priorities, the scheduling being in terms of time slots and/or frequencies at least. This is a convenient way to divide spectral resources, suitable for various widely used standards. Other ways are conceivable such as code division, and combinations of these and other ways of dividing the spectrum. Step 210 shows detecting at the MAC layer for each UE whether an indication of its radio channel condition has reached the threshold of deterioration. If yes, then at step 221 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE, in terms of a reduced number of frequencies and/or a reduced number and/or reduced size of time slots. Step 231 shows making available that shared spectral resource no longer used by that UE, for scheduling a number of frequencies and/or a number of time slots for other UEs to increase their share of the shared spectrum resource.

Fig 4 embodiment with scheduling according to buffer levels,

Figure 4 shows a variation of the embodiment of figure 2 and shows operational steps of a scheduler using the same reference numbers as figure 2 where appropriate. At step 203, there is a normal operating step of scheduling shared spectral resources to share between UEs at least according to criteria such as their priorities, the scheduling being carried out also according to demand in the form of indications of levels of data in buffers awaiting transmission in respect of any of the UEs. This is a convenient way of influencing the scheduling in combination with the influence of demand, to make more spectral resource available to others. Such indications of levels can be direct indications from downlink buffers in the base station for example, or indirect indications in the form of estimates of levels in buffers at remote locations such as at the UEs. Such estimates can be extrapolated from recent buffer status reports from UEs for example. Step 210 shows detecting at the MAC layer for each UE whether an indication of its radio channel condition has reached the threshold of deterioration. If yes, then at step 223 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE. This reduced share can be implemented in various ways, by reducing the indication of buffer level relied on by the scheduler, or by responding to a new transmission request by increasing indicated buffer level by an increment which is less than a typical increment, or by reducing a relative priority of the UE, or by reducing or blocking a dependency of the scheduling on the respective buffer level for example. Another example involving indications of buffer levels, with more detailed description is set out below in relation to figure 10. Step 230 shows making available that shared spectral resource no longer used by that UE, for scheduling for other UEs to increase their share of the shared spectrum resource.

Fig 5, embodiment with reduction also to scheduling made according to predetermined sequences and/or predicted demand

Figure 5 shows another variation of the embodiment of figure 2 and shows operational steps of a scheduler using the same reference numbers as figure 2 where appropriate. At step 204, there is a normal operating step of scheduling shared spectral resources to share between UEs at least according to criteria such as their priorities, at least some of the scheduling being carried out according to a predetermined sequence, such as for example persistent or semi persistent scheduling and/or according to an indication of predicted demand, such as for example enhanced buffer estimation. The use of a predetermined sequence can reduce the scheduling overhead involved in making regular scheduling requests for the same amount of the shared spectral resource, which can be useful where the bandwidth need or data rate does not vary unpredictably, such as a VoIP call, for example. Where an indication of predicted demand is received, an example of this is a service aware function which can predict a bandwidth needed, and influence scheduling in various ways. One way is to directly alter the appropriate buffer level indication. Step 210 shows detecting at the MAC layer for each UE whether an indication of its radio channel condition has reached the threshold of deterioration. If yes, then at step 224 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE. This can be implemented by reducing the scheduling made according to the predetermined sequence and/or adapting the sequence to cause the reduced share. For the case of predicted demand, this can be implemented by reducing the share scheduled according to the demand, or by adapting the prediction of demand to cause the reduced share for example. Step 230 shows making available that shared spectral resource no longer used by that UE, for scheduling for other UEs to increase their share of the shared spectrum resource. Figs 6 and 7, embodiment with entry to and exit from a poor channel state

Figure 6 shows a variation of the embodiment of figure 2 and shows operational steps of a scheduler using the same reference numbers as figure 2 where appropriate. At step 200, there is a normal operating step of scheduling shared spectral resources to share between UEs at least according to criteria such as their priorities. Step 210 shows detecting for each UE whether an indication of its radio channel condition has reached the threshold of deterioration. If yes, then at step 225 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE, and a poor channel state is entered for that UE. Step 230 shows making available that shared spectral resource no longer used by that UE, for scheduling a number of frequencies and/or a number of time slots for other UEs to increase their share of the shared spectrum resource. At step 240, for the UE in poor channel state, a detection is made of whether the indication of its radio channel condition has improved to reach a second threshold. If yes, for that UE the poor channel state is exited so as to return to the normal scheduling state in which this UE is allowed to increase its share of the shared spectral resource, with some hysteresis to avoid toggling between states.

Figure 7 shows a state diagram for the embodiment of figure 6 or for other embodiments. There is shown a normal scheduling state 260, and a poor channel state 270 in respect of a UE. While in the poor channel state there is reduced scheduling, in the form of reduced grants of downlink shared spectral resource, and reduced assignments of uplink shared spectral resource. The reduced scheduling may still include sufficient to allow reporting of buffer levels. To transition from the normal scheduling state to the poor channel state, there is the condition of the channel showing deterioration worse than threshold, which can be tested based on DTX and CRC indications for example.

Fig 8, embodiment with a threshold based on DTX and/or CRC

Figure 8 shows a variation of the embodiment of figure 2 and shows operational steps of a scheduler using the same reference numbers as figure 2 where appropriate. At step 200, there is a normal operating step of scheduling shared spectral resources to share between UEs at least according to criteria such as their priorities. Step 212 shows detecting for each UE whether an indication of its radio channel condition has reached the threshold of deterioration based on DTX which is an indication of discontinuous transmission, based on measurements of detected energy, such as received power. The reaching of the threshold can also be based additionally or alternatively on CRC information which indicates a rate of bit errors, which can reflect radio channel conditions. A benefit of using both is that a DTX alone may not distinguish between uplink and downlink interference, and only indirectly indicates how well data is being transmitted. CRC alone is based on data errors and so does indicate more directly how well data is transmitted, but is not specific to radio channel conditions. Further details are set out below in relation to the example of figure 10. If the threshold is reached, then at step 220 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE. Step 230 shows making available that shared spectral resource no longer used by that UE, for scheduling for other UEs to increase their share of the shared spectrum resource.

Fig 9, embodiment with reduced scheduling sufficient for detection and reporting of radio channel conditions

Figure 9 shows a variation of the embodiment of figure 2 and shows operational steps of a scheduler using the same reference numbers as figure 2 where appropriate. At step 200, there is a normal operating step of scheduling shared spectral resources to share between UEs at least according to criteria such as their priorities. Step 210 shows detecting at the MAC layer for each UE whether an indication of its radio channel condition has reached a threshold of deterioration indicating failure of transmission over a period of time. If yes, then at step 227 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE with sufficient of the shared spectral resource to enable detection and reporting of radio channel conditions. This can help enable rapid increase in scheduling when appropriate, to reduce needless latency, while still reducing wastage of the shared spectrum resource. Another example also involving a reduced share with sufficient to enable detection and reporting of radio channel condition is set out below in relation to figure 10, and described in more detail. Step 230 shows making available that shared spectral resource no longer used by that UE, for scheduling for other UEs to increase their share of the shared spectrum resource.

Fig 10, embodiment showing protocol stack with MAC layer, showing scheduler Figure 10 shows an embodiment of apparatus suitable for an LTE example. There is shown a schematic view of a base station 100 for a cell of a cellular wireless communications system, and having an eNB protocol stack 300 implemented by a processor and memory 95. Packets such as IP packets forming a data radio bearer DRB for transmission are passed down the stack, and correspondingly, received data is passed up the stack. A similar stack is provided in each UE. LTE communications are in the form of signalling radio bearers (SRB) or data radio bearers (DRB). SRBs carry control-plane data between the Radio Resource Control layers (RRC, 3GPP TS 36.331 ) 350 found at the top of the stack in the eNodeB and the UE. DRBs carry user-plane data for the end user. Each DRB is associated with a UE and a specific quality of service (QoS) - a UE may use separate DRBs for separate applications, e.g. voice and web browsing. Below the RRC layer 350 in the stack there is a PDCP layer 340. Next there is a Radio Link Control (RLC, 3GPP TS 36.322) layer, which sits above the Medium Access Control (MAC, 3GPP TS 36.321 ) layer in the eNodeB protocol stack. The MAC layer interacts with the physical layer PHY 310 below it in the stack, which converts digital transport channels in to RF signals.

Data is transferred between the MAC layers in the UE and eNodeB using transport blocks sent via shared transport channels (DL-SCH and UL-SCH). The MAC layer combines RLC logical channels, MAC Control Elements, and Hybrid-ARQ (HARQ) retransmissions into the transport blocks.

As shown in figure 10, the MAC layer includes a scheduler 1 10 which decides how to allocate time slots and frequencies for different logical channels corresponding to different UEs. Multiple logical channels may be multiplexed by multiplexer 323 for UEi, and multiplexer 324 for UEn. There may be many UEs not shown for the sake of clarity. Each of the transport channels for each UE is passed through a HARQ retransmission process 360. The scheduler 1 10 may receive inputs for each transport channel in the form of a DTX indication and a CRC indication. These may be generated from the HARQ parts. The CRC indication can be an indication of data error rate in transmission in either direction, and the DTX can be derived from a measurement of received signal power.

UL HARQ can be used to check for radio channel conditions. This can be based on detected energy in the eNB for all UL transmissions and measurements below a configured threshold can be considered DTX. A DTX indication can occur when the UE correctly received the UL grant on PDCCH, but the transmitted signal was too weak to be detected by the eNB or because the UE did not correctly receive the UL grant on PDCCH and therefore didn't transmit anything. Thus DTX does not distinguish between uplink and downlink.

In addition to the DTX measurements each UL transmission can be verified with a CRC check, for example using the HARQ parts. This can help overcome the weaknesses of DTX of being dependent on threshold configuration, and not directly indicating how successful the data transmission is. CRC alone is an indirect indicator of the radio channel condition. The combination of DTX and CRC can be used to indicate if the UE's radio connection is deteriorated so far as to indicate failure of transmission. One implementation involves counting combinations of the following:

isDTX==TRUE and CRC=NOK.

When this count exceeds a threshold over a period of time, or exceeds a rate, a poor channel state can be declared for a given UE, to cause the scheduler to change to a poor channel state.

Using the isDTX gives some indication as to whether the grant on PDCCH was detected and UL CRC indicates whether transmissions on PUSCH are working. If both indicate failure (potential failure) then there is more certainty in concluding the UE has lost coverage, compared to just depending on one of them. Note that for the uplink alone it would be sufficient to only use CRC results to detect 'failed transmissions'. One could also use DTX, though this has the disadvantage that it is set based on a threshold, so it may be incorrect in some cases. However there is clearly the possibility of combining CRC and DTX in two ways, firstly using a logical "or" of CRC and DTX, and secondly using a logical "and" of CRC and DTX.

Note that for the downlink, the UE will perform a CRC check on the transmissions it receives and send an ACK/NACK response back to the eNB. On the eNB's side this will be detected as ACK/NACK or isDTX. isDTX in downlink can be detected by checking if the HARQ feedback information ACK/NACK is received at the eNB side. This downlink isDTX can be triggered due to poor downlink channel quality with the effect that the scheduling assignment transmitted PDCCH is not decoded. Therefore the downlink count of failed transmissions can be a logical combination of the ACK/NACK and DTX results.

To implement this poor channel behavior the bufferSizes in the scheduler can be reduced or set to zero and any periodic functions e.g. Enhanced Buffer Estimation for VoLTE, that inject data into this UE's buffer estimation can be interrupted, to reduce or prevent that action, so as not to trigger the scheduling of new transmissions for this UE. Alternatively the reduction in scheduling can be implemented by blocking the dependence of the scheduling on buffer size indications.

As the UE may regain coverage at any time, optionally a periodic granting (periodicSR) of the UE in UL or DL can be performed while in the 'poor channel' state. Similarly an SR received on PUCCH can trigger an UL grant. The results of these grants shall continue to adjust the 'count' parameter and if the value of 'count' decreases below the second threshold then normal scheduling can be resumed. The frequency of the periodic grants, on the one hand can determine the delay this UE experiences before resuming normal transmissions after regaining coverage, but also impacts the number of wasted resources if this UE does not regain coverage.

An example of how to handle PUCCH Scheduling Request's. Conventionally part of the UL PUCCH channel is used for Scheduling Request (SR). An SR allocation implies a frequency allocation with a recurring periodicity (e.g. 10ms), that a UE can use to indicate to the eNB that new data has arrived in a UE's UL buffer. This shall trigger the eNB to send a grant for the UE to transmitt this data on the PUSCH channel. In a typical implementation an SR detection triggers the eNB to place X bytes of data into a selected priority queue (simulated UL buffer), which in turn will make this UE a candidate for scheduling. The selection of which priority queue to place this X bytes of data in will dictate this UE's priority against other UE's for a PUSCH allocation.

This conventional arrangement can be changed as follows. When an SR is detected for a UE which has entered the 'poor channel condition' state, this triggers an allocation of a smaller than typical value of X (X represented in some cases as an increment to an indicated level of data in buffer), smaller than the usual allocation value, and optionally also this UE receives a lower priority compared to other UE's not in this state. The smaller than typical value of X will mean spending fewer PUSCH resources in the initial grants (before receiving a Buffer Status Report (BSR) from the UE). The smaller value of X can in some cases be set to allow the UE send some user data in the initial grant e.g. a ping packet or a VoLTE packet, thus reducing latency for that data. The minimum size of X would be just enough to receive a BSR e.g. 2 Bytes.

The lower priority can give the benefit that the eNB would not spend resources on many grants on this UE before it gets some feedback on the first grant. The eNB must anyway prioritise at least one grant as answer to the SR, but the situation to avoid is the combination of large increment X that can result in many grants and a high priority causing the UE having a poor channel condition to be selected often over those other UE's which can send data more successfully. This shows that the smaller than typical X is beneficial in reducing wastage of shared spectral resource.

Other embodiments can be applied to systems other than LTE. One example is WCDMA MAC. Among the main differences is that WCDMA still uses some circuit switching, but LTE uses entirely packet switching. Also, for LTE the MAC is centralized so that the eNB has full control of the transmission but in WCDMA the UL MAC is partly distributed.

Fig 1 1 , embodiment with MAC layer scheduling of LTE logical channels system Fig 1 1 shows operational steps according to another embodiment with MAC layer scheduling of logical channels of an LTE system. It shows a variation of the embodiment of figure 2 and shows operational steps of a scheduler using the same reference numbers as figure 2 where appropriate. At step 208, the normal operating step of scheduling shared spectral resources involves scheduling in the form of centralized mac scheduling of logical channels of the LTE system by making grants of time slots and frequencies for downlink logical channels onto a downlink shared transport channel and making assignments of time slots and frequencies for uplink logical channels onto an uplink shared transport channel. Step 210 shows detecting for each UE whether an indication of its radio channel condition has reached the threshold of deterioration indicating failure of transmission and therefore representing inefficient use. If the threshold is reached, then at step 228 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE, with fewer grants for downlink logical channels and/or fewer assignments for uplink logical channels. This can involve scheduling smaller transport block sizes. The transport block size can be determined by number of frequencies and modulation and coding scheme. Step 230 shows making available that shared spectral resource no longer used by that UE, for scheduling for other UEs to increase their share of the shared spectrum resource.

Fig 12, state diagram with enhanced buffer estimation

Figure 12 shows a state diagram showing a talk state 500 and a silent state 510 for a VoIP call example, according to an embodiment. In the talk state, the scheduling includes enhanced buffer estimation based on predicted demand for bandwidth when the VoIP call is transmitting talking. In the silent state, the call is kept open, but the bandwidth needed is less, so the scheduling includes enhanced buffer estimation based on predicted demand for bandwidth when the VoIP call is transmitting no talking. There is also shown a poor channel state 520, which is entered when the DTX and CRC show channel conditions worse than a threshold. In this state the scheduling has reduced amounts of grants and assignments and the enhanced buffer estimation is stopped or blocked. In the condition that the DTX and CRC show channel conditions better than a second threshold, the poor channel state is exited. Not shown, only for the sake of clarity, is the poor channel state being entered from the talk state and returning to the talk state.

Similiarly SemiPersistent Scheduling can be turned off. Those scheduling opportunities can then be reused for other UE's.

Figure 13, embodiment with reduced share for worst performing of the UEs when cell overloaded

Figure 13 shows a variation of the embodiment of figure 2 and shows operational steps of a scheduler using the same reference numbers as figure 2 where appropriate. At step 200, there is a normal operating step of scheduling shared spectral resources to share between UEs at least according to criteria such as their priorities. Step 210 shows detecting at the MAC layer for each UE whether an indication of its radio channel condition has reached the threshold of deterioration. If yes, then at step 220 the scheduling is reduced to provide a reduced share of shared spectral resource for that UE. Step 230 shows making available that shared spectral resource no longer used by that UE, for other UEs to increase their share of the shared spectrum resource. At step 255, there is a step of detecting whether the cell is overloaded. This can be carried out in various ways, for example involving assessing whether a desired quality of service is being met for all UEs, or by determining a total capacity in use for example. If the cell is regarded as overloaded, and if other mechanisms to handle this at higher protocol levels are too slow to react, then the overload can be relieved by reducing a share of the worst performing of the radio channels. This can be achieved by determining which is the worst performing radio channel based on recent measurements, and causing it to reach the threshold to cause a reduced share to be scheduled, by adapting the threshold accordingly, during the overload. The situation can be reassessed periodically, and the threshold reduced again once the overload is not present, or if the selected radio channel improves and is no longer the worst performing.

In one example the eNB can hold a level of overload per cell based on violation levels of QoS for the connected UE's. The overload levels can be based on a combination of

the degree of violation for an individual UE e.g. how much time are packets delayed over the QoS delay budget

The number of UE's suffering violation.

The threshold for 'deterioration of channel quality' can be adjusted depending of the level of cell overload. This definition of overload could as well be applied to cell or eNB level, depending on which UE's are included / how they are grouped.

An alternative mechanism for detecting overload is to measure usage levels of critical resources e.g.

- cell resources e.g.

- how many Scheduling Blocks are used on the data channels (PUSCH & PDSCH)

- how many CCE's are used on the control channel PDCCH

- eNB capacity (combination of all cells) e.g.

- how many scheduling entities

- how many Scheduling Blocks

- throughput

Comparing the usage levels of these resources with thresholds can indicate levels of load that can in turn be used to adjust the threshold.

Concluding remarks

Even a small number of UEs losing coverage through poor radio channel condition, while having data in buffers which are highly prioritized by the scheduler, can negatively impact the capacity of a cell / eNB. This negative impact continues until either the scheduler's buffer estimates are brought to zero or these UEs are released by other supervision functions. Embodiments can reduce this impact by providing a quicker detection of UE's with poor channel conditions, compared to the reaction time of existing functions. This detection can then be used to restrict their impact and increase cell / eNB throughput and/or reduce latency for other UE's in that cell / eNB. Other variations and additions can be envisaged within the claims.

Claims

1 . A method of scheduling for a cell of a cellular radio communication system serving mobile user equipments, UEs, the method having steps of:

scheduling a shared spectral resource to share it between the mobile

UEs,

detecting at a media access control protocol layer for at least one of the mobile UEs whether an indication of its radio channel condition has reached a threshold of deterioration indicating failure of transmission over a period of time, and

if so, scheduling a reduced share of the shared spectral resource for the said at least one of the mobile UEs without waiting for release from the cell of the said at least one of the mobile UEs, and making at least some of the shared spectral resource no longer scheduled for the said at least one mobile UE, available for others of the mobile UEs.

2. The method of claim 1 , the step of scheduling shared spectral resource comprising scheduling time slots and frequencies, and the reduced share comprising at least one of: fewer of the time slots, a reduced size of the time slots and fewer of the frequencies.

3. The method of any preceding claim, the step of scheduling with a reduced share comprising scheduling sufficient of the shared spectral resource for detection and reporting of the radio channel condition for the said at least one of the mobile UEs.

4. The method of claim 1 , 2 or 3, the step of scheduling shared spectral resource comprising scheduling according to an indication of level of data in buffers awaiting transmission, and the step of scheduling a reduced share comprising at least one of: reducing the indicated level of data in buffers in respect of the said at least one UE to cause the reduced share, responding to a new request for transmission by increasing the indication of level of data by less than a typical increment, reducing a priority of the data in buffers relative to that of the others of the UEs, and reducing a dependency of the share for the said at least one UE on its respective indicated level of data in buffers.

5. The method of any preceding claim, the step of detecting when the threshold has been reached comprising detecting whether the cell is overloaded and adapting the threshold if the cell is overloaded, to cause a worst performing one of the radio channels to reach the threshold of deterioration.

6. The method of any preceding claim, the step of scheduling shared spectral resource comprises scheduling according to at least one of: a predetermined sequence, and a predicted demand, and the step of scheduling the reduced share comprising at least one of: reducing the share scheduled according to the predetermined sequence, adapting the sequence, reducing the share scheduled according to the predicted demand, and adapting the prediction of demand.

7. The method of any preceding claim, having a step of operating in a poor channel state according to the indication of the radio channel condition being worse than the threshold, and having the step of exiting the poor channel state according to an indication of the radio channel being better than a second threshold, with hysteresis provided between the entry into and the exiting from the poor channel state.

8. The method of any preceding claim, the detection step being based on at least one of: detected energy received over the radio channel, and a result of a CRC step.

9. The method of any preceding claim, the scheduling comprising centralized medium access control scheduling of logical channels of an LTE system and the step of reducing comprising reducing the scheduling of at least one of: assignments of time slots and frequencies for downlink logical channels onto a downlink shared transport channel and grants of time slots and frequencies for uplink logical channels onto an uplink shared transport channel.

10. A computer program on a non transitory computer readable medium and having instructions which when executed by a computer, cause the computer to carry out the method of any of claims 1 to 9.

1 1 . Base station apparatus for a cell of a cellular radio communication system for serving mobile user equipments, UEs, the base station apparatus comprising: processing means operative to:

schedule a shared spectral resource to share it between the mobile UEs, detect at a media access control protocol layer, for at least one of the mobile UEs, whether an indication of its radio channel condition has reached a threshold of deterioration indicating failure of transmission over a period of time, and if so, schedule a reduced share of the shared spectral resource for that one of the mobile UEs without waiting for release from the cell of the corresponding mobile UE, and to make at least some of the shared spectral resource no longer scheduled for that mobile UE, available for others of the mobile UEs.

12. The base station apparatus of claim 1 1 , the processing means comprising a processor and a memory, said memory containing instructions executable by said processor to carry out the scheduling.

13. The base station apparatus of claim 1 1 or 12, the processing means being operative such that the scheduling of the shared spectral resource comprises scheduling time slots and frequencies, and the reduced share comprises at least one of: fewer of the time slots, a reduced size of the time slots and fewer of the frequencies.

14. The base station apparatus of any of claims 1 1 to 13, the processing means being operative to carry out the scheduling with a reduced share while scheduling sufficient of the shared spectral resource for detection and reporting of the radio channel condition for the said at least one of the mobile UEs.

15. The base station apparatus of any of claims 1 1 to 14, the processing means being operative such that the scheduling of shared spectral resource comprises scheduling according to an indication of level of data in buffers awaiting transmission, and the step of scheduling a reduced share comprises at least one of: reducing the indicated level of data in buffers in respect of the said at least one UE to cause the reduced share, responding to a new request for transmission by increasing the indication of level of data by less than a typical increment, reducing a priority of the data in buffers relative to that of the others of the UEs, and reducing a dependency of the share for that UE on its respective indicated level of data in buffers.

16. The base station apparatus of any of claims 1 1 to 15, the processing means being operative to detect when the threshold of deterioration has been reached by detecting whether the cell is overloaded and being operative to adapt the threshold if the cell is overloaded, to cause a worst performing one of the radio channels to reach the threshold of deterioration.

17. The base station apparatus of any of claims 1 1 to 16, the processor being operative to schedule the shared spectral resource by scheduling according to at least one of: a predetermined sequence, and a predicted demand, and

the processor also being operative to schedule the reduced share by at least one of: reducing the share scheduled according to the predetermined sequence, adapting the sequence, adapting the share scheduled according to the predicted demand, and adapting the prediction of demand.

18. The base station apparatus of any of claims 1 1 to 17, the processing means being operative in a poor channel state according to the indication of the radio channel condition being worse than the threshold, and being operative to exit the poor channel state according to an indication of the radio channel condition being better than a second threshold, with hysteresis provided between the entry into and the exiting from the poor channel state.