Reverse Link Rate Control in 1xEV-DO with Adaptive Antenna Arrays

Abstract

Rate control and distributed reverse link scheduling in CDMA systems are considered. Probabilistic rate allocation with and without token bucket constraints gives mobiles more autonomy in making transmission decisions. A system with adaptive antenna arrays deployed at the base stations is proposed as a generalization to the model with omnidirectional antennas. Uniform linear antenna arrays with a microstrip patch as their constituent element are presented. Their number of elements is varied and their performance is compared to the omnidirectional case, showing substantial throughput gains. Index Terms-Antenna arrays, distributed reverse link scheduling, rate control, 1xEV-DO, CDMA I. INTRODUCTIONIn recent years, distributed scheduling of mobile users has gained more importance. Probabilistic transmission by mobile devices in the EV-DO system is described in [1], whereas the incorporation of token buckets in the EV-DO Rev A system is discussed in [2]. Consequently, mobile devices are given more autonomy in making transmission decisions, and thus more freedom of action ([3], [4], and [5]). This differs from the traditional aspects of wireless networks, where the base stations maintain full and centralized control in the allocation of the system resources.In centralized schemes (e.g. maximum C/I [6] and proportional fair [7]), user scheduling is the responsibility of the base station, which normally allocates resources to a single user at a given time instant. Distributed schemes are allowed by several standards for modern 3G/4G cellular networks, e.g., 1xEV-DO Rev A [8]. These standards have introduced mechanisms that give devices greater independence in making transmission decisions best matched to their applications, e.g. deciding when to transmit and at what rate. The price for this flexibility is potentially higher interference, and a corresponding degradation in performance [3], [4]. Consequently, base stations should maintain some control over the interference. They exercise this control by setting a Reverse-link Activity Bit (RAB) on when the load increases, and clearing it otherwise. The RAB is used in conjunction with a token bucket, where tokens are a function of transmission power, and the base station controls the token generation rate and the token bucket depth [3]. The reverse traffic channel rate control algorithm for cdma2000 is presented in [1], where transition from one rate to another is performed by mobile devices in a probabilistic manner based on the value of the RAB. The reverse traffic channel MAC design of cdma2000 1xEV-DO Rev A system is discussed in [2], where an algorithm is presented to update the token generation rate and the token bucket depth based on the computation of the RAB. In this work, we generalize