The 18th Annual IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC’07) A MICRO-ECONOMICS APPROACH FOR SCHEDULING IN CDMA NETWORKS WITH END-TO-END QOS GUARANTEES Walid Saad American University of Beirut, Department of Electrical and Computer Engineering Beirut, Lebanon, Email: wes02@aub.edu.lb Sanaa Sharafeddine Lebanese American University, Division of Computer Science and Mathematics Beirut, Lebanon, Email: sanaa.sharafeddine@lau.edu.lb Zaher Dawy American University of Beirut, Department of Electrical and Computer Engineering Beirut, Lebanon, Email: zaher.dawy@aub.edu.lb ABSTRACT Third generation CDMA networks strive to deliver high speed data services through a shared radio channel with scarce re- sources. To efficiently utilize the available radio resources, we propose a new scheduling algorithm based on techniques from micro-economics. Unlike existing literature that mainly focuses on maximizing total system and/or individual user util- ity, this new algorithm aims at ensuring QoS guarantees from end to end for all active connections. Moreover, it considers the time varying channel conditions in both uplink and downlink directions jointly rather than each direction separately. Simula- tion results show that the proposed algorithm allows several users to simultaneously transmit while providing end-to-end QoS guarantees in terms of frame success rate and end-to-end delay. I. I NTRODUCTION The main driver behind the deployment of next generation CDMA-based cellular networks is their ability to provide mo- bile stations (MSs) with innovative multimedia services that demand high resources and impose stringent QoS require- ments. One way to efficiently exploit the system resources is to use advanced scheduling algorithms. This resource alloca- tion problem has been recently a major concern for a number of research papers; the majority of which sought to find the best scheduling algorithm in either the uplink or the downlink. This paper proposes a novel scheduling algorithm for CDMA networks. The key distinctive features of this algorithm are three-fold: i. it schedules the users in communicating pairs based on the time-varying conditions of both the uplink and the downlink channels jointly, ii. its main allocation criterion is to provide end-to-end QoS guarantees, and iii. it is based on techniques taken from micro-economics such as indifference curves and marginal rate of substitution. The paper is organized as follows. Section II presents a sur- vey of related literature. Section III describes the system model and Section IV presents the proposed scheduling algorithm. Simulation results are presented and analyzed in Section V. Fi- nally, conclusions are drawn in Section VI. II. LITERATURE SURVEY Early work in the area of uplink scheduling was done in [1] for a network having only voice users. The authors provide a power allocation algorithm that minimizes the total uplink transmitted power ensuring that every user achieves its target SNR. The authors in [2] and [3] propose an uplink scheduling algorithm for a CDMA network having a single data service. The uplink power control game is modelled as a non cooper- ative game where every MS tries to maximize its utility func- tion. Also, jointly with the scheduling algorithm, [3] considers a base station (BS) assignment algorithm that helps in meeting the objective of maximizing individual utilities. Several next generation services are asymmetric and highly resource demanding in the downlink direction. Hence, exhaus- tive research activities focused on finding resource allocation algorithms suitable for the downlink. For instance, a joint BS and resource allocation algorithm is presented in [4] that maximizes the total utility (with price) per individual BS. This paper borrows the utility function used for the uplink in [3] and applies it to the downlink. The BS assignment scheme is modelled as a non cooperative game, and the output of the pricing based power and rate allocation problem is used in or- der to assign some MSs in highly loaded BSs to other lightly loaded BSs. In [5] and [6] the downlink resource allocation problem for a multi-cell heterogenous CDMA system is con- sidered. In contrast to previous work, both papers take into account the rather non-critical delay requirements of these ser- vices. The power allocation problem is modelled using the Multiple Choice Multi Dimensional Knapsack (MMKP) prob- lem. The main goal is to jointly allocate power and assign users to different BSs in a manner that maximizes total system utility. The research done in [7] builds on the works of [5] and [6] in order to provide a new approach to the scheduling problem. The proposed algorithm allocates the total available transmit power of the BS to one MS during the transmission of a frame within a time slot. Moreover, unlike previous work, the utility function is defined per packet and not per user. The authors in [8] make yet another contribution in the downlink resource allocation problem. The aim of the paper is to find a joint BS assignment and power allocation algorithm that maximizes throughput and fairness in a CDMA system. 1-4244-1144-0/07/$25.00 c 2007 IEEE