WO2008104045A1

WO2008104045A1 - Method of allocating communications resources and scheduler therefore

Info

Publication number: WO2008104045A1
Application number: PCT/CA2007/000317
Authority: WO
Inventors: Thierry De Corte; Donald Stevenson; Rouzbeh Khatibi
Original assignee: Wavesat Inc.
Priority date: 2007-02-28
Filing date: 2007-02-28
Publication date: 2008-09-04
Also published as: BRPI0721380A2

Abstract

The present invention relates to a method for allocating communications resources amongst a plurality of connections and a scheduler therefore. The method includes a step of evaluating needs of each connection, and summing the needs of each connection to obtain total needs for all connections. Then the method pursues weighting the needs of each connection with corresponding ones of the total needs. Afterwards, the method allocates the communication resources amongst the plurality of connections in accordance with the weighted needs. In turn, the scheduler includes an information module, a calculating module and an assignment module. The information module identifies needs of each one of a plurality of connections. The calculating module performs the summing of the needs and the weighting of the needs of each connection, while the assignment module allocates communication resources amongst the plurality of connection in accordance with the weighted needs.

Description

METHOD OF ALLOCATING COMMUNICATIONS RESOURCES AND

SCHEDULER THEREFORE

FIELD OF THE INVENTION

The present invention relates to the allocating of communications resources amongst a plurality of connections, and more particularly to a method and scheduler for allocating communications resources amongst a plurality of connections taking under consideration some quality of service (QoS) requirements.

BACKGROUND OF THE INVENTION

Nowadays, telecommunications networks are widely used to transport large amount of packets and run multiple connections with different QoS requirements simultaneously, and sometimes for one subscriber. To meet such demands, Institute of Electrical and Electronics Engineers (IEEE) 802.16 standard defines granting of bandwidth on a per subscriber basis, as opposed to a per connection basis. The bandwidth grants are determined and broadcast by base stations to subscriber stations through UL-MAP messages.

It is however a well-known problem to efficiently and fairly distribute the granted bandwidth amongst multiple connections. Multiple prior art publications have proposed some partial solutions to the latter issue. One of those publications is an IEEE publication titled "Quality of Service support in IEEE 802.16 networks", written by Claudio Cicconetti et al., presented in the IEEE Network, March/April 2006. This publication presents mechanisms available in IEEE 802.16 networks to support QoS. More particularly, the publication relates simulations performed in two distinct scenarios. The simulations demonstrate the effectiveness of the 802.16 Medium Access Control (MAC) providing differentiated services to applications with different QoS requirements, such as Voice over Internet Protocol (VoIP) and Web. The publication also reports the differences in delay between downlink and uplink channels, and report higher delays and sharper increase in delay in the uplink channels.

Another publication of interest, titled "Efficient Fair Queuing Using Deficit Round-Robin", written by Shreedar et al., presented in the IEEE/ACM Transactions on networking, volume 4, no. 3, of June 1996, describes a mechanism to improve the efficiency and fairness in queuing of multiple flows over allocated network resources. For doing so, the authors propose use of stochastic fair queuing for assigning flows to queues. Then, for servicing the flows, a round-robin mechanism is used, and tracking of deficits is performed. While this solution improves the fairness between multiple queues, it does not take under consideration additional requirements imposed by QoS.

In the publication titled "Packet Scheduling for QoS support in IEEE 802.16 broadband wireless access systems" written by Wongthavarawat et al., published in the International Journal of Communication Systems, 2003, volume 16, pages 81-96, describes a scheduling algorithm for IEEE 802.16 standard, providing QoS support to different traffic classes. For doing so, each type of traffic class is handled differently. The algorithm starts by applying overall bandwidth allocation following strict priority, and policing of maximum bandwidth by traffic class. Then, a second level of granularity of bandwidth allocation is performed on a connection type level. Although this publication provides a solution to the issue of bandwidth allocation, the latter does not resolve overall fair and simple bandwidth allocation. Quite to the contrary, this publication provides a solution in which higher priority traffic classes are serviced first, and left bandwidth is used for the remaining traffic classes.

There is thus a need for a scheduler and a method for allocating uplink communication resources simply and fairly, with quality of service.

SUMMARY OF THE INVENTION The present invention relates to a scheduler and a method for allocating network resources amongst a plurality of connections in a simple and fair manner. Contrary to prior art solutions, the present solution is simple, and thus requires less computational capabilities than other prior art solutions, while achieving fairness amongst multiple connections while respecting quality of service.

In accordance with a first aspect, the present invention relates to a method of allocating communication resources amongst a plurality of connections. The method evaluates needs of each connection, and sums the needs of each connection to obtain total needs for all connections. The method then weights the needs of each connection with corresponding ones of the total needs and allocates the communication resources amongst the plurality of connections in accordance with the weighted needs.

In accordance with another aspect, the present invention relates to a scheduler. The scheduler includes an information module, a calculating module and an assignment module. The information module identifies needs of each one of a plurality of connections. The calculating module sums the needs of each connection to obtain total needs for all connections. The calculating module further weights the needs of each connection with corresponding ones of the total needs. The assignment module allocates communication resources amongst the plurality of connections in accordance with the weighted needs.

BRIEF DESCRIPTION OF THE DRAWINGS The following drawings will be used in conjunction with the following Detailed Description of the invention, to describe various aspects of the present invention, in which:

Figure 1 depicts an exemplary flowchart for a subscriber station Unicast Grant in accordance with IEEE 802.16 standard; Figure 2 is a flowchart of a method in accordance with an embodiment of the present invention;

Figures 3a and b are flowcharts of a method in accordance with another embodiment of the present invention; and Figure 4 is a schematic scheduler in accordance with an embodiment of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to data communications, and more particularly to data communications of a type similar to IEEE 802.16, in which uplink bandwidth is allocated on a per subscriber basis, instead of on a per connection basis. For allowing optimized management of the bandwidth, subscriber stations (including stationary and mobile equipment, for example wireless computers, wireless Personal Digital Assistants, and the like) are equipped with a scheduler. A scheduler is a component of the subscriber station that allocates communication resources, such as uplink bandwidth, over ongoing connections. In the present context, a connection refers to a path between two terminals, such as the subscriber station at one end, and a base station at another end. As raised in many prior art documents, the allocation of communication resources is a delicate task, with many consequences. An improper allocation of resources may result in some connections having good quality of service, and other connections having barely any service. Furthermore, many known allocation algorithms require extensive processing capabilities, which may result in increased power consumption, and thus not suitable for power sensitive and processing limited applications, such as wireless applications.

One of the additional particularities of data communications such as the IEEE 802.16 is the fact that the base station, on a needed basis, performs the bandwidth allocation dynamically. The need is based on a perception at the base station of the subscriber requirements, the defined quality of service for each established connection when available, and the available uplink bandwidth. Thus the uplink bandwidth allocated to each subscriber is not known in advance, and dynamically modified by the base station. The uplink bandwidth allocated by the base station is communicated to the subscriber station through UL-MAP (acronym used to represent a set of information that defines the entire access for a scheduling interval) messages. The UL-MAP are broadcast from the base station all subscriber stations and define an uplink bandwidth usage interval. The uplink bandwidth usage interval may be assigned to either a unicast, a multicast or broadcast address. When the subscriber station receives the grant size, it then must perform some choices as to how the grant should be allocated over the various connections, task which is typically performed by the scheduler, (hereinafter called a grant size).

Reference is now made to Figure 1 , which depicts an exemplary flowchart for handling by a subscriber station of a Unicast Grant in accordance with IEEE 802.16 standard. Upon receipt of the UL-MAP, the latter is first processed (step 110). Then, verification is made in step 120 of whether the received UL-MAP, whether the received UL-MAP contains a grant(s) assigned to the subscriber's Basic Connection Identifier (CID). If the grant is for Basic CID, bandwidth is assigned in step 130 and data is sent over the uplink connections. Additionally, step 140 may also be used to send bandwidth request messages to the base station, to request additional bandwidth.

In the context of the present invention, a first embodiment of the method for allocating the communications resources (including more particularly uplink bandwidth) of the present invention is shown on Figure 2. The method generally includes 4 main steps: a step of evaluating needs (step 210), a step of summing the needs (step 220), a step of weighting the needs (step 230) and a step of allocating the resources (step 240). More particularly, the method of the present invention aims at optimizing the allocation of the grant size, while respecting committed quality of service of all connections. Examples of quality of service parameters may include Minimum Reserved Traffic Rate (MRTR), Maximum Latency (ML), and all other quality of service parameters commonly known for connections. The MRTR refers to a minimum amount of data to be transported on behalf of the connection, averaged over time. MRTR thus represents the bandwidth to be guaranteed to the connection even in the advent of a heavily loaded network. The measure of the data packets for MRTR considers payload size at an input of a Medium Access Layer (MAC), without MAC overhead and optional CRC field (Cyclic Redundancy Check). The ML specifies a maximum amount of time a data packet is allowed to wait in a queue of a connection, before being transmitted. The ML quality of service preferably shall be met only when the connection is transmitting below its MRTR value. Other quality of service parameters could also be considered in the context of the present invention, but for ease of understanding, the following description will only refer to the MRTR and ML. Thus, upon receipt of the UL-MAP message, the method starts with evaluating the needs in step 210. The evaluation of the needs may take various forms. For example, the needs may include only number of data packets needing to be transmitted to meet quality of service parameters (such as MRTR and/or ML), or a combination of quality of service parameters and a number of packets in a queue for each one of the connections. More particularly, in accordance with an embodiment of the present invention, the evaluating of the needs is performed by calculating a deficit (step 250) and calculating a surplus (step 260) for each of the connections. Various equations can be used to calculate the deficit of each connection. Such equations may include: (a) a maximum value of the MRTR and the number of packets in the queue exceeding the predefined ML criterion for the connection;

(b) comparing the calculated deficit value with the number of packets in queue for the connection and selecting a lesser one as the calculated deficit value; and

(c) ensuring that the calculated deficit value is larger than a Bandwidth Request grant header.

The method pursues with the step 260 for calculating the surplus value for each of the connections. The surplus value may consist for example of a difference between the number of packets in queue for that connection and the calculated deficit value.

The step 220 of summing the needs includes, in accordance with an embodiment of the present invention, two sub-steps of summing the deficits of all connections (step 270), and summing the surplus of all connections (step 280), so as to obtain an overview of the total needs for all connections.

The method then pursues with the step of weighting the needs in step 230. That step may be performed by dividing each of the needs of each one of the connections by a corresponding one of the total needs. It is thus possible afterwards to proceed with the allocating of the uplink bandwidth to multiple connections in accordance with the weighted needs (also called calculated share hereinbelow). By measuring the needs of each connection, and summing those needs, it is thus possible to allocate the uplink bandwidth to the various connections in a manner that is fair, simple, and respect quality of service requirements.

Reference is now made concurrently to Figures 3a and 3b, which depict flowcharts of a method in accordance with another embodiment of the present invention. The method starts in step 305 with a step of initializing values for a total deficit, a total surplus and a total bandwidth request (total_bw_req) to 0. Then the method proceeds for each of the connection (identified as "i") in the following manner: (a) determine a number of data packets in a queue for that connection (step 310);

(b) if the number of data packets in the queue for that connection is above 0, proceed to step 320, or else, proceed to step 325;

(c) obtain the MRTR value for the connection in step 320;

(d) if the MRTR value for the connection is above 0, proceed to step 330, or else, proceed to step 335;

(e) in step 330, determine a number of data packets exceeding the predefined ML for the connection;

(f) in step 335, set the number of data packets exceeding the predefined ML for the connection to 0;

(g) then, in step 340, the deficit value is calculated in the following manner: deficit = max( MRTR, late); deficit = min (number of data packets in the queue, deficit); deficit = max (sizeof( bw-req header) deficit);

(h) additionally in step 340, the surplus value is calculated in the following manner: surplus = # of packets in queue - deficit value;

(i) in step 345, the calculated deficit value and surplus value for the connection are added to the total deficit value, total surplus value, and total bandwidth request value; (j) and in the event that there was no packet in the queue, the method proceeded to step 325, where the MRTR value, the late value, the deficit value and surplus value for the connection were all set to 0.

Then, the method continues in Figure 3b, where a first verification is performed in step 350 to determine whether the total deficit is greater than the allocated bandwidth (grant). If the total deficit is greater than the grant, the method proceeds with step 355, while if the total deficit is smaller than the grant, the method proceeds with step 360. The remaining of the method is performed on a per connection basis.

In step 355, the deficit value for the connection is compared with 0. If the deficit value for the connection is greater than 0, the method proceeds with step 365, otherwise it is terminated. In step 365, the allocated bandwidth (also called share) of the granted bandwidth is determined in the following manner: share = sizeof (bw-req header) + (grant - total_bw_req)* (deficit/total deficit); share = min (deficit, share); share = max (sizeof(bw_req header, share). The method then continues in step 370 by deducing from the total bandwidth request value the size of the band_req header.

Alternatively, in the event that the value of the total deficit is not greater than the grant, the method proceeds with step 360, where a determination is made on the number of packets in the queue. In the event that the number of packets in the queue for the connection is greater than 0, the method continues with step 375, otherwise, it is terminated.

In step 375, the allocated bandwidth for the connection is calculated in the following manner: share = deficit + (grant - total deficit)^*(surplus/total surplus); share = min (total, share); share = max( sizeof (bw-req header), share).

The method then continues in step 380 with recalculation of the total deficit and the total surplus. Finally, in step 385, the value of the grant is also recalculated to deduct the share allocated to the connection.

Referring now to Figure 4, there is depicted a schematic scheduler 400 in accordance with an embodiment of the present invention. The scheduler 400 of the present invention is mainly composed of an information module 410, a calculating module 420 and an assignment module 430. The information module 410 identifies needs of each one of the pluralities of connections. The needs can be determined by comparing stored agreed quality of service criteria per connection and collected status information for each of the connections. Examples of quality of service criteria may include MRTR, ML, etc..) The status information includes the number of packets in queue for the connection. Additional status information could be obtained for example by communicating 440 with a Traffic Policer to obtain transmission rate of each uplink connections. The information module 410 may also obtain information on the delay undergone by each packet in the queues by comparing a current time with a time-stamp applied on every packet upon entering a MAC layer.

The results of the information module are then communicated to the calculating module 420. The calculating module 420 sums the needs of each connection to obtain total needs for all connections. The calculating module 420 further calculates the deficit value and surplus value for each connection. The calculating module 420 further sums the deficit values and surplus values of all connections to obtain the total deficit value and the total surplus value. The calculating module 420 further weights the needs of each connection with corresponding total needs to obtain a value of a share of bandwidth for each connection. Then, the results of the calculating module 420 are communicated to the assignment module 430, which task is to allocate the granted bandwidth to each of the connection in accordance with the calculated share. [0001] The information module 410, calculating module 420 and assignment module 430 could respectively consist of hardware material, mounted on a board, or incorporated within on chipset, or alternatively, be implemented in the form of a software.

The present invention has been described by way of preferred embodiment. It should be clear to those skilled in the art that the described preferred embodiments are for exemplary purposes only, and should not be interpreted to limit the scope of the present invention. The method and scheduler as described in the description of preferred embodiments can be modified without departing from the scope of the present invention. The scope of the present invention should be defined by reference to the appended claims, which clearly delimit the protection sought.

Claims

CLAIMS:

1. Method of allocating communication resources amongst a plurality of connections, the method comprising steps of:

evaluating needs of each connection;

summing the needs of each connection to obtain total needs for all connections;

weighting the needs of each connection with corresponding ones of the total needs; and

allocating the communication resources amongst the plurality of connections in accordance with the weighted needs.

2. The method of claim 1 , wherein the needs include a number of packets in queue for the connection being evaluated, and a Minimum Reserved Traffic Rate (MRTR) for the connection.

3. The method of claim 2, wherein the needs further include a number of packets in the queue exceeding a predefined Maximum Latency (ML) criterion.

4. The method of claim 3, wherein the step of weighting the needs is performed by dividing the needs of each of the connection by the total needs.

5. The method of claim 3, wherein the step of evaluating needs includes:

calculating a deficit value for the connection, the deficit value being equal to a maximum value of the MRTR and the number of packets in the queue exceeding the predefined ML criterion for the connection.

6. The method of claim 5, wherein the step of calculating the deficit value further includes comparing the calculated deficit value with the number of packets in queue for the connection and selecting a lesser one as the calculated deficit value.

7. The method of claim 6, wherein the step of calculating the deficit value further ensures that the calculated deficit value is larger than a Bandwidth Request grant header.

8. The method of claim 7, further including a step of calculating a surplus value for each of the connections, the surplus value being equal to a difference between the number of packets in queue for that connection and the calculated deficit value.

9. The method of claim 8, wherein the step of summing the needs of each connection to obtain total needs for all connections includes summing the calculated deficit and surplus of all connections to obtain total deficit and total surplus.

10. A scheduler comprising:

an information module for identifying needs of each one of a plurality of connections;

a calculating module for summing the needs of each connection to obtain total needs for all connections and for weighting the needs of each connection with corresponding ones of the total needs; and

an assignment module for allocating communication resources amongst the plurality of connections in accordance with the weighted needs.

11. The scheduler of claim 10, wherein the identifying of the needs is based on: stored agreed quality of service criteria per connection and collected status information for each of the connections.

12. The scheduler of claim 11 , wherein the needs include a number of packets in queue for the connection being evaluated, and a Minimum

Reserved Traffic Rate (MRTR) for the connection.

13. The scheduler of claim 12, wherein the needs further include a number of packets in the queue exceeding a predefined Maximum Latency (ML) criterion.

14. The scheduler of claim 13, wherein the weighting of the needs is performed by: dividing the needs of each of the connection by the total needs.

15. The scheduler of claim 14, wherein the calculating module further calculates a deficit value for each of the connections, the deficit value being equal to a maximum value of the MRTR and the number of packets in the queue exceeding the predefined ML criterion for the connection.

16. The scheduler of claim 15, wherein the calculating of the deficit value further includes comparing the calculated deficit value with the number of packets in queue for the connection and selecting a lesser one as the calculated deficit value.

17. The scheduler of claim 16, wherein the calculating the deficit value further includes ensuring that the calculated deficit value is larger than a Bandwidth Request grant header.

18. The scheduler of claim 17, wherein the calculating module further calculates a surplus value for each of the connections, the surplus value being equal to a difference between the number of packets in queue for that connection and the calculated deficit value for that connection.

19. The scheduler of claim 18, wherein the calculating module further sums the calculated deficit and surplus of all connections to obtain a total deficit and a total surplus.