WO2009080081A1

WO2009080081A1 - A scheduling method and system for communication

Info

Publication number: WO2009080081A1
Application number: PCT/EP2007/011351
Authority: WO
Inventors: Matteo Andreozzi; Andrea Bacioccola; Daniele Franceschini; Ettore Pulieri; Roberto Rossi; Giovanni Stea
Original assignee: Telecom Italia S.P.A.
Priority date: 2007-12-21
Filing date: 2007-12-21
Publication date: 2009-07-02
Also published as: BRPI0722339B8; EP2223478A1; EP2223478B1; CN101926133B; BRPI0722339B1; US20100278152A1; US8908511B2; CN101926133A; BRPI0722339A2

Abstract

In a method of scheduling transmission services over a high speed packet access communication link such as a HSDPA and/or a HSUPA link a list of queues is created (200) for transmission services to be provided over the link. The queues, which include both Real Time and Non Real Time queues, are allotted respective service priorities based e.g. on a channel quality indicator and/or a Quality of Service indicator to produce an ordered list of queues based on the service priorities. The link resources needed for serving at least one set of queues having higher priorities in the ordered list are estimated (206) and a check is made (208). If these resources are available, the set of queues having higher priorities in the ordered list are served (210, 216, 218). If the resources required are not available, at least one queue is removed (214) from the ordered list of queues.

Description

A SCHEDULING METHOD AND SYSTEM FOR COMMUNICATION NETWORKS; CORRESPONDING DEVICES, NETWORK AND COMPUTER PROGRAM PRODUCT

Field of the invention

The present disclosure relates to scheduling techniques for communication networks.

This disclosure was devised with special attention paid to its possible use in scheduling transmission of data to and/or from mobile stations in a cellular communication system whose radio access network exploits a high speed Code Division Multiple Access (CDMA) technology, e.g. HSDPA (High Speed Downlink Packet Access) for downlink and/or HSUPA (High Speed Uplink Packet Access) for uplink. Reference to this preferred field of application is not however to be construed in a limiting sense of the scope of this disclosure.

Description of the related art

Document WO2007017753 addresses the aspect of fast scheduling in a radio base station configured for HSDPA operation and proposes a solution to the problem of low latency tolerant services as discussed by Bang Wang et al . in the conference paper "VoIP Performance in HSDPA Simulation Study", VTC 2005 Spring, 30 May-1 Jun. 2005, Stockholm, Sweden. In this arrangement the selectable users are ordered by relying on criteria known in the art such as Proportional Fair or Round Robin, and thus proposes to consider a user as eligible if at least one of the following conditions are satisfied: a minimum number of stored packets,- a maximum delay time; and any pending retransmission packets.

The solutions suggested by the paper by Bang Wang et al. address the problem of optimizing spectral efficiency of HSDPA for low bit rates services, at the cost of introducing higher latency delays. In fact, the paper by Bang Wang et al . proposes to equally share power and codes between the selected users . . The arrangement disclosed in WO-A-2006/081570 uses a combination of two essential metrics in order to schedule allocation of channel resources.

The former metric is related to the time- to-expiration of packet (s) for a user stored in the scheduling buffer and the second metric relates to the throughput that the user can get during the scheduling interval. Time-to-expiration is a metric that reflects the amount of time left before the packet would be considered useless. Two options are used in order to prioritize packet transmission to maintain QoS and efficiently utilize the bandwidth available on a multi-user system. In both options, a two-part performance metric is used in a procedure corresponding to the following basic mathematical equation: Priority = Delay Metric + Rate Metric. In the first option, the first metric is a delay metric corresponding to the number of packets with a time till expiration less than a designated threshold. The latter metric is a rate metric and corresponds to the number of complete packets that the current packet payload can accommodate if that user is chosen. For the second option, the first metric is a delay metric computed by summing a fractional computation derived using the time till expiration of all the packets in the queue for a user. The second metric is a rate metric and corresponds to the number of complete packets that the current packet payload can accommodate if that user is chosen.

The document US-A-2006/0153216 describes a scheduler for a downlink packet data traffic channel shared by a plurality of mobile stations by ranking at least some of the mobile stations based on delay. The scheduler employs a ranking metric for a mobile station that varies directly with the mobile station scheduling downlink transmission rate and a delay factor indicative of the staleness of the data queued for the mobile station. Calculating the ranking metric may comprise calculating the ranking metric as a function of the scheduling rate divided by the difference between a threshold and the delay factor. The threshold may represent the maximum allowed delay, and the delay factor may be based on an amount of elapsed time associated with the oldest of the data queued for the mobile station. The delay factor may be based on an amount of buffer consumed by the queued data. Queued data is advantageously transmitted to the mobile station based on the scheduling in a multi-user downlink packet on the shared downlink packet data channel . Such an approach allows for delay sensitive data, such as VoIP data, to be scheduled with increased urgency when quality of service is about to be compromised.

Object and summary of the invention

Starting from the preceding discussion, the Applicant has noted that the solutions considered in the foregoing fail to provide a satisfactory response to a number of needs , such as , e.g.: properly exploiting a class of service as a parameter to assign the priority to data transmission, while also distinguishing between retransmissions and transmissions to be able to prioritize the former over the latter and taking into account the possibility to multiplex multiple users in one Time Transmission Interval, or TTI, at least for the downlink shared channel; optimizing the scheduling procedure of e.g. a

HSDPA/HSUPA radio link mainly in order to render performance for Real Time users acceptable in terms of

Quality of Service perceived by the users and in terms of capacity of the system;

- providing low latency delay links whilst reducing the spectral efficiency by an amount as small as possible, with a view to optimizing performance in respect of Real Time (RT) users (e.g. in terms of Quality of Service or QoS as perceived by the users and in terms of capacity of the system) as well as Non Real Time (NRT) users, thus providing an exhaustive scheduling procedure that also takes into account Non-Real Time users;

- properly prioritizing RT users by taking into account - on the one side - the channel condition as experienced by the user and - on the other side - the delay accumulated by the data of the user,-

- determining (i.e. calculating) the power needed to transmit an amount of data as a function of the Transport Block Size or TBS selected, with the objective of minimizing the chance of retransmission while making an efficient use of this limited resource.

The object of the invention is thus to provide a satisfactory response to those needs. According to the present invention, that object is achieved by means of a method having the features set forth in the claims that follow. The invention also relates to a corresponding system (i.e. a scheduler), corresponding communication devices including such a scheduler, a related network as well as a related computer program product, loadable in the memory of at least one computer and including software code portions for performing the steps of the method of the invention when the product is run on a computer. As used herein, reference to such a computer program product is intended to be equivalent to reference to a computer-readable medium containing instructions for controlling a computer system to coordinate the performance of the method of the invention. Reference to "at least one computer" is evidently intended to highlight the possibility for the present invention to be implemented in a distributed/ modular fashion.

The claims are an integral part of the disclosure of the invention provided herein.

An embodiment of the arrangement described herein is thus mainly focused on defining the criteria to allocate radio resources over a packet switched network where a High Speed technology is implemented in the radio access network of the telecommunication system to provide both Real Time (RT) and Non Real Time (NRT) services.

As used herein, "High Speed" technology will be generally associated with the deployment of a HSDPA (High Speed Downlink Packet Access) system over the downlink and/or the deployment of a HSUPA (High Speed Uplink Packet Access) over the uplink in radio access.

In certain embodiments, the main technology issue for considering both the downlink and the uplink may relate, on the one side, to the strict requirements of Real Time service in terms of Quality of Service perceived by the user and, on the other side, to the nature of the High Speed technology that uses shared radio resources to serve multiple users in both the two links: in particular, in downlink, HSDPA allocates power and codes between multiple users while in uplink HSUPA indirectly decides the impact that a user has on the interference perceived by the system by allocating the power for the user's transmission. High Speed technology may serve all the packet switched users, including those of the NRT services. In a High Speed embodiment, decisions may be taken in respect of the priority order of the users and the radio resources allocation criteria in order to serve multiple traffic classes with such flexibility as to satisfy every flow of data, belonging to either a RT or a NRT service.

In an embodiment, a high-level architecture of a scheduling procedure is defined which, on the one hand, optimizes the performance of the RT service over the packet switched network and, on the other hand, is flexible in such a way as to satisfy also NRT, considering both uplink performance (as implemented in the radio base station for allocating the uplink resources of the HSUPA radio access technology) and downlink performance (as implemented for allocating the downlink resources of the HSDPA radio access technology) .

In an embodiment, based on knowledge of the resources, i.e. power and codes, required to transmit the unsent packets to the UE addressed the resources are shared between all the selected users by taking into account the requirements associated with each selected user.

An embodiment of a high-level architecture in HSDPA and in HSUPA may involve the definition and usage of technology specific parameters to detail the HSDPA downlink scheduling procedure and the HSUPA uplink scheduling procedure, as described in the following.

Exemplary principles which may be applied in such an embodiment may include one or more of the following: retransmission such as H-ARQ (Hybrid Automatic Repeat reQuest) retransmissions may have priority over transmissions; the same radio resources may be scheduled to retransmit data in terms of power and number of channelization codes used in the first transmission by the NodeB, when considering the downlink, and the power assigned by the NodeB to the UE, when considering the uplink;

- in first time transmissions, the RT traffic may have priority over NRT traffic; in other words, the scheduler may consider the NRT traffic only when no other RT traffic can be served;

- the RT and NRT queues may be sorted according to a priority function; in particular, new transmissions of RT data may be based on a priority function derived by combining various metrics: for instance, a metric may be related to the channel quality observed by the user; a metric may be related to the time to expiration of packets for the user; a metric may be related to the previous one and express the amount of data whose expiration time is the most urgent. New transmission of NRT data may be prioritized based on one metric that considers the amount of data available to be sent for each service. The uplink priority function and the downlink priority function may use the same metrics; however, the implementation of these metrics may exploit technology specific parameters;

- the RT traffic may use as few radio resources as possible thanks to a capacity estimation function, i.e. the amount of resources assigned for transmission of RT packets is limited to the quantity strictly needed to achieve the required (BLock Error Rate) BLER target of the service. The downlink capacity estimation function and the uplink capacity estimation function may use the same criteria but the implementation may exploit technology specific parameters ;

NRT traffic may be given all the remaining resources, i.e. those not used by RT flows; - NRT traffic flows may be scheduled so as to maximize system usage.

The arrangement described herein fully meets the needs outlined in the foregoing. For instance, it may exploit class of service as a parameter to assign the priority to data transmission, and distinguish between retransmissions and transmissions to prioritize the former over the latter and taking into account the possibility to multiplex multiple users in one Time Token Interval or TTI.

Also, it may optimize the scheduling procedure of e.g. a HSDPA/HSUPA radio link in order to render performance for

Real Time users over HSDPA/HSUPA technology acceptable in terms of Quality of Service as perceived by the users and in terms of capacity of the system.

The arrangement described herein may provide low latency delay links whilst reducing the spectral efficiency by as small as an amount as possible, thus optimizing performance of Real Time (RT) users as well as Non Real Time (NRT) users.

Also, it may properly prioritize RT users by taking into account both the channel condition as experienced by the user and the delay accumulated by the data of the user.

Additionally it may determine the power needed to transmit an amount of data as a function of the TBS selected.

Brief description of the annexed drawings

The invention will now be described, by way of example only, with reference to the enclosed figures of drawing, wherein:

Figure 1 is a pictorial diagram generally representative of the context of this disclosure; Figure 2 is a high level state diagram of a scheduler architecture as disclosed herein;

- Figure 3 is a flow chart representative of various functions included in the scheduler architecture as disclosed herein; and

Figure 4 is a flow chart representative of a residual resource allocation function for use in exemplary HSDPA and HSUPA scheduling procedures as disclosed herein.

Detailed description of exemplary embodiments

In the following description, numerous specific details are given to provide a thorough understanding of embodiments . The embodiments can be practiced without one or more of the specific details, or with other methods, components, materials, etc. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of the embodiments .

Reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments .

The headings provided herein are for convenience only and do not interpret the scope or meaning of the embodiments .

As indicated, figure 1 is a pictorial diagram generally representative of the context of this disclosure, and thus refers to a mobile network arrangement wherein one or more base stations (Node B) serve a plurality of user equipments . Only one base station BS and two user equipments, designated UEl, UE2, are shown for the sake of simplicity and without loss of generality.

In an exemplary 3G (or over) embodiment supporting High Speed (e.g. HSDPA/HSUPA) operation, the base station BS may include specific High Speed (HS) MAC functionalities such as MAC-hs in downlink for HSDPA and MAC-e in uplink for HSUPA plus additional functionalities such as packet scheduling, link adaptation, and H-ARQ. Transmission in the

"downlink" direction (i.e. BS to UE) will include signaling

(e.g. Channel Quality Indicator or CQI, Ack/Nack) and data, while transmission in the "uplink" direction (i.e. UE to BS) will include signaling (e.g. scheduling information, Ack/Nack, Transport Block Size) , and data.

In the following, some exemplary aspects of scheduling for High Speed technology are formalized and a set of definitions and notations that will be used in the rest of this disclosure are provided.

The notations used in this disclosure are reported in Table 1, with the proviso that parameters/acronyms related to HSDPA technology are well known in the art and do not require to be explicitly specified; conversely, parameters/acronyms related to HSUPA technology are explicitly specified for immediate reference. Subscripts are used to denote a User Equipment (UE) , while a functional notation is used to denote a time, which is expressed in multiples of the TTI (Time Token Interval) . Table 1 NOTATION

Scheduling (both downlink and uplink) will generally mean selectively allotting usage of the communication (i.e. "service") resources to a plurality of users (or "queues") both real time (RT) and non real time (NRT) .

More to the point, a scheduler in the base station BS may perform the following functions: - handling services priority (i.e. RT and NRT);

- scheduling new transmissions and retransmissions; and

- allocating the radio resources, i.e. transmission power and channelization codes for HSDPA and transmission power for HSUPA, to assign to each UE.

For RT services one may assume that the discard timer in the Radio Link Control (RLC) protocol layer is configured in order to drop stale data. The definition "stale data" applies to those packets that would certainly miss a relevant deadline of theirs. These may be, i.e. those packets whose delay will be in excess of the Quality of Service delay requirement imposed by the nature of the service: for instance, the ultimate deadline for a conversational service may be 400 ms. Figure 2 is a high level state diagram of a scheduler architecture as disclosed herein.

At the beginning of TTI j , either or both of the two schedulers (downlink and uplink) are invoked and the following operations are performed. Hereinafter an exemplary condition will be considered wherein both schedulers (i.e. HSDPA and HSUPA) are invoked. State 100 (Schedule H-ARQ reTx)

H-ARQ retransmissions are scheduled first, i.e. they are given the highest priority. According to the chase combining H-ARQ mode, the configuration of the physical layer may remain the same in order to retransmit the same data packet as in the original transmission. This allows the receiver to combine the energy perceived per bit in the receiver buffer where the original packet has been stored.

The HSDPA technology may mandate that the number of codes employed in the H-ARQ process remain constant, whereas the power assigned to each transmission can be dynamically changed. In fact, the exemplary scheduler disclosed herein serves each H-ARQ retransmission with the same codes and power as in the original transmission. The HSUPA technology may instead assign the same Absolute Grant as in the original transmission, i.e. the same amount of power that was originally assigned in the first transmission.

Reference 102 is representative of the scheduler remaining in state 100 if there are H-ARQ users to be scheduled.

State 104 (Schedule new transmissions)

If there are resources available, a set of new UEs are scheduled to receive new data. The available resources are shared among these new UEs, so that each queue is given the minimum resources required for servicing its most urgent

Packet Data Units (PDUs) .

Reference 106 is representative of the scheduler remaining in state 104 if there are enough resources to schedule one new transmission.

State 108 (Allocate residual resources)

This step is performed if and only if the resources have not been completely depleted in the previous steps. In HSDPA the remaining codes and power are shared among the UEs selected at state/step 104, if any.

Otherwise, the remaining power is shared among the set of H-ARQ retransmissions, which may imply that not all codes are actually used.

In HSUPA the remaining interference perceived by the

Node B is used by allocating power to the UEs selected at point 2, otherwise between the UEs selected at state/step

102. In fact, assigning resources for uplink may mean configuring one UE to use a given power level in uplink transmission by correspondingly generating interference at the base station or BS. The BS will thus have a maximum threshold of interference to be perceived.

This obviously maximizes the chance of successful transmission in both the cases.

State 110 (Trigger data to the physical layer) In HSDPA this means that the MAC-hs PDUs are created and the each MAC-hs PDU is transmitted over the HS-DSCH. In HSUPA the B_ed_i(j) are sent to the UEs in order to make each UE build the MAC-es/e PDUs that will be transmitted over the relevant E-DCH. As better detailed in figure 3, the following operations are performed when scheduling new transmissions (state/step 104) .

In a step 200, eligible users are selected. The set of new transmissions to be scheduled are selected every TTI among the set of eligible users.

In HSDPA a user is marked as eligible in TTI j when it satisfies the following set of conditions:

UE₁ HfARQ(J) VUnACKeCi(J)

m 3k : B_tJ[ ≥ 0 , i , . e „ . , ^ the UE >₁ has backlog ; j rpillaasslt ,, r rpp/ITTTTJl < •

¹I ^{+ 1}I — J Note that In HSUPA the UE communicates its backlog to the radio base station in the Total E-DCH buffer status (TEBS) field included in the Scheduling Information (SI) .

In a step 202 a check is made as to whether the list of eligible users/queues is empty. In the case of a positive outcome of step 202, the function terminates, and in a step 203 the list of scheduled transmissions is passed to the next block.

If steps 202 yields a negative outcome, in a step 204 the scheduler sorts the list of eligible users according to a priority function: the list of eligible users may be sorted according to a priority function, which may be formally defined in the "Utility Function" section. The priority function may use technology specific parameters and therefore the implementation may be different in the downlink and in the uplink even though both the HSDPA and HSUPA priority functions are based on the same metrics.

New transmissions of RT data may be based on the priority built over the combination of three metrics. A first metric may be related to the channel quality observed by the user to which the queue is associated. The second metric may relate to the time to expiration of packets for the user. The third metric may be related to the second and express the amount of data whose expiration time is the most urgent .

New transmission of NRT data may be prioritized based on one metric that considers the amount of data that the user would be able to send divided by the amount of data in the queue buffer. In a step 206 the scheduler computes the resources for the highest-priority PDUs which were not accommodated yet.

The user JTF> at the head of the list is selected, and its , . , . . m = max(k:B,_t > 0) highest priority non-empty queue ^{l •} ' is considered. The number of data with the highest priority ι-Pⁿ\i^£ι,m) _{^ for} v^hich resources have not been allocated yet (initially the head-of-line PDU(s)), is determined and the resources required to serve them all are computed.

This step can be performed several times for the same user. To that purpose, a set of utility functions may be defined referring to the Capacity Estimation function, which are formally described below. This capacity estimation function may use technology specific parameters; therefore the implementation may be different in the downlink for HSDPA and in the uplink for the HSUPA system.

In a step 208, the scheduler checks if there are enough resources. To that effect, the scheduler analyzes the output of step 206 and determines if all the Service Data Units (SDUs) can be transmitted in the current TTI. If there are enough resources to transmit all the SDUs computed in the step 206 (i.e. positive outcome of step 208), the execution continues to step 210. Otherwise (i.e. negative outcome of step 208) , step 212 is performed.

"Enough resources" means, in the HSDPA system, that the number of codes and the power computed by the capacity estimation function are smaller than or equal to the number of codes and the power reserved to HSDPA. In the HSUPA system, this will mean that the total interference that the radio base station (i.e. the receiver side of the uplink) will experience in the next TTI, i.e. the Received Total Wideband Power (RTWP) , computed by the capacity estimation function, does not exceed the RTWP_MAX selected by higher layers .

In the step 212, the scheduler will reduce the number of SDUs considered. The scheduler may use the utility functions defined below in order to compute the maximum number of SDUs that can be transmitted in the forthcoming TTI, according to the amount of available resources.

When this happens, in a step 214 ■ is removed from the list of eligible users and its scheduled SDUs, if any, will be transmitted in the forthcoming TTI. As an outcome of this step,

may be found to be unable to transmit anything given its current channel state.

Whatever the step previously performed (i.e. step 208 or step 214, as a function of the outcome of step 208) in the step 210 the scheduler will update the list of scheduled users .

If in steps 206, 208, 212 a new user is scheduled for which at least one SDU has been selected, the latter is added to the set of scheduled users for the forthcoming TTI

^V' . The available resources are updated to reflect the new allocation.

In a HSDPA system this means that the number of available codes and available power destined for HSDPA UEs are updated, while in the HSUPA system the total received interference still available compared to RTWP_MAX value is updated in the radio base station.

In a step 216, the scheduler will then check if the new total number of users does not exceed the maximum number of users per TTI or there are no more available resources. If either of these conditions is met (positive outcome of step 216) , then the user selection process is complete and in a step 218 the eligible list is set equal to "O) . Otherwise (negative outcome of step 216) , there might still be room for more traffic, and therefore the scheduler goes back to step 202 after updating the resources available in a step 220.

The above function returns the set of new PDUs that will be served in the forthcoming TTI for each user, together with the required radio resources .

Figure 4 is a block diagram of the allocate residual resources function (state 108 of figure 2) which can be developed in both the HSDPA and HSUPA scheduling procedure.

This function is invoked at the end of scheduling new transmissions (state 104 of figure 2) , if there are residual resources .

Essentially, this procedure involves a step 300 wherein the scheduler gets the list of retransmissions and the list of new transmissions. In as step 302 a check is made as to whether the list of new transmissions is empty.

If that is the case (positive outcome of step 302) , then in a step 304 the scheduler allocates resources to the H-ARQ retransmissions. In the case of a negative outcome of step 302, which means that the list of new transmissions is not empty, then in a step 306 the scheduler allocates resources to the new transmissions .

In either case the scheduler then evolves towards the state 110 of figure 2.

The utility functions that the scheduler uses are described in the following.

Queue sorting Exemplary priorities for real time (RT) and non real time (NRT) queues are now described.

With regard to RT traffic, as far as VoIP (Voice over IP) is concerned, a maximum one-way mouth-to-ear delay may be 400 ms. This implies that a given "delay budget" ^DTv^έi*⁾

(henceforth referred to as deadline) is set on each VoIP packet for its transit through the UMTS domain. More specifically, the cell capacity is observed to decrease as the deadlines become tighter.

In an embodiment, priorities are assigned according to an Hybrid Channel-Aware & Real-Time (HY-CART) policy. "Hybrid" because it behaves as both a MAX-C/I and a pure Earliest Deadline First (EDF) approach, thus keeping into account both the channel state and the packet deadlines simultaneously. Furthermore, the relative importance of the two contributions changes as the deadline approaches, so as to minimize the chance of a deadline miss.

In a pure MAX-C/I approach, the priority of a RT queue in the downlink may be computed as follows:

where i is the index of the selected user. This method assigns the highest priority to the UE that currently experiences the best channel conditions . The channel conditions are estimated as described previously, in order to draw a fair comparison among UEs that reported CQI values in different times and power conditions. Although such an index may can be used in order to maximize throughput, such a policy may turn out to be unfair, since it may lead to repeated deadline misses for UEs who experience non optimal channel conditions, even at moderate cell loads.

On the other hand, a pure EDF scheme computes the priority of a queue as :

. 1 P^{n> =} DT(Q₁^)-J This gives a higher priority to packets with the shortest residual lifetime; this ratio goes to infinite and becomes negative when queue i has missed a deadline. In this case, the head-of-line packet is dropped by the scheduler.

The HY-CART approach blends the above two approaches, as it aims at influencing the RT traffic priority based on both the deadline of the data involved in the RT communications and the channel quality. More specifically, when there are RT queues to be served and deadlines are "far enough", we strive to optimize the system performance, serving the UEs with better channel conditions (which implies using less resources, possibly scheduling more users simultaneously, and making more codes and power available for lower priority transmissions) .

In HY-CART, priorities may be computed as follows: pri, = a_t ^■ ChQuality + (\-a_i )-k hi_ pri{Q_{i k} ) where ^aι is a dynamic weight, O≤α≤l, and * is a positive scaling factor that will differ from HSDPA and

HSUPA, which is required in order to make the two addenda comparable. While the first term is self-explanatory, the second term reflects the number of bits of those packets in the queue having the most stringent deadline (assuming that the channel conditions remain the same throughout the TTI interval) . The weight ^α< is a function of ^(Qu^{<> }} _t i.e. it accounts for the deadlines. When the latter is large (i.e. the head-of-line packet of a queue is far from its deadline), ^a< is equal to 1, which makes the second term null. On the other hand, ^aι approaches zero as ^(βι^⁾-^J approaches zero, so as to make the priority strictly depending on deadlines. At least notionally, there are many ways to devise a function ^a'^~J\ ^'-^k' ^J) . por instance, the values proposed in Table 2 may be chosen.

Table 1. ^ai reference values

Discard Time of the head-of-line PDU in "^ik '

DT(Q._k) > T₄ i

T₃<DT(Q_ik)<T₄ %

T₂ < DT(Q₁J <T₃ *

DT(Q._k) < T₂ 0

In Table 2 , ^i , *i _an(j T_{4 are} three thresholds chosen considering VoIP requirements in terms of one-way-delay.

The queue sorting function may use technology specific parameter to develop the first metric mentioned in the formula above, in particular: ■ HSDPA: ^CAβ«^ = C^β/

- HSUPA: ^ChQ→ =^UPH

With regard to NRT traffic, policies aimed at achieving fairness are currently being investigated. NRT traffic is also directly influenced by the TCP behavior. In fact, the throughput of TCP depends on the RTT of the connection.

Exemplary HSDPA and HSUPA approaches will now be described. Throughout this final section of the description current (e.g. ETSI) standard definitions and acronyms will be used. The meaning of these standard definitions and acronyms is well known to those of skill in the area and do not require detailed explanations herein. As a HSDPA approach the following may be used: . TBS₁

This function gives a higher priority to users which require less TTIs in order to clear their backlog. As a HSUPA approach the following may be used:

TBS hsupa i pn, = = ^Z-^

TEBS₁

This function gives a higher priority to users which require less TTIs in order to clear their backlog.

Now, a Transport Control Protocol (TCP) flow can have a low backlog for essentially two reasons: - the application that the TCP carries is about to pause, in which case delaying the TCP packets would delay the completion of the ongoing transaction; the TCP has a small congestion window, which prevents it from using more bandwidth. However, in this case, the TCP whose bandwidth has the lowest resource cost should be favoured, so as to increase the TCP rate.

As the TCP rate increases, the size of the queue is going to increase as well. Thus, this strategy strives to tune the TCP rates so that each source takes the same amount of resources. In doing this, it makes a fair usage of the resources .

Capacity Estimation

In the case of HSDPA, reference is made to the Compute CQI-Power Table.

The ^^Ji is the feedback sent from UE i to the

NodeB at a TTI ^rePι^<J , J being the TTI for which a scheduling decision is required. Each CQI is associated with a specific modulation and coding scheme, a specific number of codes and a corresponding TBS . The UE reports the highest CQI value which would make a TBS worth of bytes be received with a BLER no greater than 10%.

The UE computes the CQI based on the total received HS-PDSCH power, i.e. on P^HSPDSCH = p^CPI™ + r + Δ , where: - Δ, the reference power adjustment, depends on the CQI mapping table defined for each UE category,

- F, the measurement power offset, is signaled by higher layers and configured by the RRC.

UE power allocation is related to both the

and the CQI selected for the next transmission. However, the power used from the UE to compute the CQI at TTI ^rep< might not 'be the same the UE will see at TTI j .

Thus, a method is devised to compute the power that is assigned to the UE, while guaranteeing a BLER no larger than 10%.

P-" I Λ

The fading ^{■ KJJ} is estimated by means of a function that inverts the CQI reporting procedure as follows:

_{CQi rVona}ι J SNR₁ (X) _{CQ Oβsa}

SMUx) =P^ (X)-If (X) ₁ * '' [ [ CCQQII""""^10* * * J J _{{ 0 1 )} where

if"' (x) = 10 -log₁₀ - lθ ¹⁰ + 10 ¹⁰

and ^a , ^^intra , ^^inte , and P¹"" are supposed to be known and it is possible that the radio base station derive them throughout the UE measurements such as RSCP and RSSI.

However, the HSDPA power

, on which (0.1) is computed at TTI x, may in fact be divided among all active users. Therefore, the power actually used for UE i, ^pΛ^x> _t pused / \ is generally smaller than (or at least equal to) - _' . Furthermore, it is likely to be different from ^Pl^^J' for ^j>x . The scheduler then estimates ' '*' as follows: pat, φ _ QQJoffie, J

According the previous formulae, the scheduler is made aware of the UE channel state. Therefore, it is able to compute for each CQI value c, the power

that should make the BLER no greater than 10%, under the hypothesis t_hat rWzPT¹U⁾.

Finally, the UE' s CQI-Power table is computed as follows:

A function designated Compute Maximum CQI computes the maximum CQI that can be assigned to the UE i according to the power that is currently available. In order to do this, it performs the "inverse" computation as the one described in the previous section, i.e. it tries to match the

IpHSDP A _ pused / -\\ available power ^ ^ '> with the related CQI in the CQI-Power table.

In the case of HSUPA, a Partial Received Interference Computation function may be resorted to. Interference may be comprised of different contributions such as the Thermal Noise that we consider as known, the Intracell Interference that is due to the users of the current cell and the Intercell Interference that is due to the users of the other cells.

This function sets up a correct starting received signal power value which will be used by the Grant Computation and Assignment function in order to check the current interference level at NodeB against the selected grant, and therefore the E-DPDCH power, for the Ue currently under consideration.

This function estimates the Intracell Interference contribution using the estimated value of the path loss for each eligible selected UE derived by the RSCP measurement, and computes a partial interference contribution for each

UE by adding an DPCCH_RX * PLOSS contribution to the current total received signal power (RTWP) . The Intercell

Interference contribution is then estimated by multiplying the estimation of the Intracell Interference for a interference factor that is an absolute value . For example the interference factor could be considered as 0.55.

If other channels are active (as example, High Speed Downlink Control Channel, HSDPCCH) they are also taken into account in the RTWP estimation.

A Grant Computation and Assignment function computes the optimum grant to be assigned to the Ue being considered.

This may involve a number of steps . A first step involves estimating the signal attenuation and compute the relevant signal power. The scheduler estimates the total path loss (Ploss) for the selected UE thanks to the RSCP and, knowing the received DPCCH power level, computes: - DPCCH transmission power : DPCCH received power over estimated path loss, i.e.:

DPCCH_TX= DPCCH_RX * PLOSS

- E-DPDCH maximum hypothetical power: according to the last reported UPH, the scheduler may compute the maximum usable power for the E-DPDCH, i.e.: E-DPDCHMAX = UPH * DPCCH_TX

Then the maximum TBS is computed. Based on the information retrieved in the previous step, the scheduler computes, the maximum hypothetical usable TBS if an Absolute Grant equal to E-DPDCH_MAX/DPCCH was assigned to this UE. This is done by looking up a TBS table with an entry value equal to B_ed-

Knowing the reference values, the look-up table may be obtained by the following equation:

This TBS is scaled down to the HLBS reported by the UE in order to eventually restrict the maximum TBS to the one strictly necessary to serve the highest priority buffer in the next TTI.

Then the scheduler estimates the transmission success probability by computing the transmission success probability for the selected UE by calculating its Ec/N0 thanks to the E-DPDCH and DPCCH powers as determined previously and thanks to the Interference previously estimated by the Partial Received Interference Computation function.

This value is then used to do a lookup in a dedicated BLER vs Ec/N0 curves for the chosen TBS . I f the resulting BLER is found to be acceptable (i.e., less than 10%) , the Grant already computed is returned for this UE.

Otherwise, given a configured threshold, the TBS is scaled down to the nearest one in the TBS selection table.

Then a new grant is retrieved from this table, which is the minimum needed to transmit at the newly selected TBS. Using this new grant value, a new E-DPDCH_MAX is computed as

GRANT*DPCCHχχ . In this case, the step is re-executed with this newly selected

value. This step ends if either a successful Grant is selected or the minimum TBS is reached, whichever occurs first

The returned grant is used to estimate a maximum E- DPDCH power which will be used in transmission by the current selected UE. This power, added to the current Total

Interference value computed by the related Utility Function after the sorting phase of the scheduling procedure, is used to compute the hypothetical RTWP and Cell Load which would be experienced by NodeB in the next TTI.

If this Cell Load value is found to be acceptable

(i.e. lower than a pre-configured threshold value), the

Absolute Grant value computed previously is confirmed and the current Total Interference is increased with the corresponding E-DPDCH hypothetical next TTI received power.

Without prejudice to the underlying principles of the invention, the details and the embodiments may vary, even appreciably, with reference to what has been described by way of example only, without departing from the scope of the invention as defined by the annexed claims .

Claims

1. A method of scheduling transmission services provided over a high speed packet access wireless communication link, the method including the steps of:

- creating (200) a list of queues for transmission services to be provided over said link, wherein said services include both Real Time and Non Real Time services, allotting (204) to said queues in said list respective service priorities based on at least one of a channel quality indicator and a Quality . of Service indicator to produce an ordered list of queues based on said service priorities, estimating (206) the link resources needed for serving at least one set of queues having higher priorities in said ordered list,

- checking (208) whether said resources are available, and:

- i) if said resources are available, serving (210, 216, 218) said at least one set of queues having higher priorities in said ordered list, and

- ii) if said resources are not available, removing (214) from said ordered list at least one queue from said at least one set of queues having higher priorities .

2. The method of claim 1, wherein Real Time service queues in said list are allotted (204) said respective service priorities based on a set of metrics including: - the channel quality observed by the user associated with the queue; the time to expiration of packets for the user; - the amount of data whose expiration time is the most urgent .

3. The method of either of claims 1 or 2, wherein Non Real Time service queues in said list are allotted (204) said respective service priorities based on a metric dictated by the amount of data that the user associated with the queue would be able to send divided by the amount of data in the queue buffer.

4. The method of any of the previous claims, including the steps of :

- providing a retransmission mechanism of packets over said wireless link, - scheduling first (100) service of said packets to be retransmitted .

5. The method of claim 4, wherein: said wireless link is a code division multiple access link having a number of channelization codes and a power reserved thereto which identify a given transport block size,

- said retransmission mechanism of packets over said link serves the packets retransmitted with the same transport block size as in the original transmission.

6. The method of any of the previous claims, wherein: said wireless link is a code division multiple access high speed downlink having a number of channelization codes and a power reserved thereto, said estimating (206) the link resources needed involves estimating a number of channelization codes and a power for serving said at least one set of queues having higher priorities in said ordered list, said checking (208) whether said resources are available involves checking whether the number of codes and the power estimated (206) are smaller than or equal to the number of channelization codes and the power reserved to said high speed downlink.

7. The method of claim 4, wherein: - said wireless link is a code division multiple access high speed downlink having an associated channel quality indicator,

- said retransmission mechanism of packets over said downlink serves the packets retransmitted with the same channel quality indicator as in the original transmission.

8. The method of any of the previous claims, wherein: said wireless link is a code division multiple access high speed uplink exposed to interference, - said estimating (206) the link resources needed involves estimating a total interference to be experienced on said uplink, said checking (208) whether said resources are available involves checking whether said total interference estimated does not exceed an upper threshold.

9. The method of claim 4, wherein: said wireless link is a code division multiple access high speed uplink, and - said retransmission mechanism of packets over said uplink assigns to the packets retransmitted the same absolute grant retransmitted as in the original transmission .

10. The method of any of the previous claims, including:

- at the end of scheduling (104) said transmission services, checking (302) if residual service resources remain available, and if residual service resources remain available, allocating (304, 306) said residual resources to further transmission services.

11. The method of claim 10, including the steps of:

- providing a retransmission mechanism of packets over said link,

- if said residual service resources remain available, in the absence of new transmission services, allocating

(306) said residual resources to retransmitting packets over said.

12. The method of any of the previous claims, wherein said link is selected out of a HSDPA and a HSUPA link, and wherein said channel quality indicator is selected out of CQI, TBS, UPH and TEBS.

13. A scheduler for scheduling transmission services provided over a high speed packet access communication link, the scheduler being configured for performing the method of any of claims 1 to 12.

14. A communication device including a scheduler according to claim 13.

15. The communication device of claim 14, in the form of a base station (BS) in a communication network.

16. The communication device of claim 14, in the form of a user equipment (UEl, UE2) in a communication network.

17. A communication network including at least one communication device according to either of claims 15 or 16.

18. A computer program product loadable in the memory of at least one computer and including software code portions for performing the steps of the method of any of claims 1 to 12 when the product is run on a computer.