Flow-level stability and performance of channel-aware priority-based schedulers

Abstract

Abstract-Channel-aware scheduling in modern wireless networks enables the system to exploit the random rate variations across different users to increase the performance of the system. We analyze channel-aware priority-based downlink scheduling policies at the so-called flow level with a stochastically varying number of users. The priority can be any monotonously increasing function of the instantaneous rate of the user, which generalizes the well-known linear weight-based policies. Also, ties are allowed within a user class, as well as between user classes. As the main result, we characterize when these priority-based policies are stable under an intuitive necessary condition, which holds for arbitrary tie breaking rules and is independent of the flow size distribution. Additionally, for the policies for which the necessary condition is not sufficient, a more stringent condition is derived in the case of two traffic classes. Finally, extensive simulations have been performed to compare the performance of different priority-based and utility-based policies. I. INTRODUCTIONOne key aspect in modern cellular systems affecting the performance is the scheduling of the radio resources among the users. In these systems, time is slotted and scheduling decisions can be made at the time scale of milliseconds. The fast scheduling and channel quality feedback enables the system to optimize the scheduling for data traffic by exploiting the fading phenomena causing random variations in the channel quality across the users. These channel-aware (also known as opportunistic) schedulers aim at increasing the throughput of the system by favoring those users having instantaneously good channels. A well-known example is the proportionally fair (PF) scheduler proposed for 1xEV-DO systems, see [1].Channel-aware schedulers can be broadly classified as utility/maxweight based and rate-based priority schedulers. Utility/maxweight based approaches use instantaneous rate information coupled with knowledge of the throughput/queue sizes/packet delays, see [2], [3], [4]. The rate-based approaches, such as the relatively best (RB) scheduler [5] and its generalizations to weight-based strategies [6], only use information about the rate (channel) statistics. Typically, these channel-aware schedulers have been analyzed at the time slot or the packet level with a static population of users, either assuming a saturated set of users, see [5], [4] or allowing packet level-dynamics and analyzing the system at the heavy traffic limit [2], [7]. The results have established that simple stochastic gradient policies of the utility and so-called maxweight policies (of either packet delays or queue lengths) have many desirable properties. However, in the above models the common aspect is that the number of users stays constant,

Radio-Aware Service-Level Scheduling to Minimize Downlink Traffic Delay Through Mobile Edge Computing

“…The new scheduler is aimed at implementing context-aware traffic shaping to minimize the average delay of traffic flows; meanwhile, eNodeB scheduling is driven by a classical implementable scheduling policy [6]. Section 2 also discusses some technological issues that determine the achievable performance in this kind of multi-user mobile service provisioning, such as the partially observable channel problem.…”

“…Undoubtedly, the most studied resource allocation problems aim at minimizing the mean transfer delay of user flows. Although several channel-aware strategies exist with the objective of minimizing the mean flow delay, Best CQI and Proportional Fair [6] among the most popular ones, the achievement of the optimal solution for time-varying scheduling optimization problems is computationally and analytically unfeasible. To cope with this problem, flow-level channel-aware opportunistic scheduling in time-varying systems has been analyzed by approximate techniques [8,9] in order to design simple, tractable and implementable well-performing heuristic priority scheduling rules [9,10,7].…”

Performance Analysis of Scheduling Algorithms for Web QoE Optimization in Wireless Networks

“…Referring to opportunistic scheduling, several channel-aware policies have been investigated in the literature, among the most known Best Rate and Proportional Fair [1]. However, flow-level models with time-varying service rates are extremely difficult to solve analytically, and to the best of our knowledge there is no mathematical resolution.…”

“…The time-varying nature of wireless channel provides an opportunity to schedule flows when they see favorable instantaneous channel conditions, which is referred to as opportunistic or channel-aware scheduling [1]. Mobile users send this channel quality information to the scheduler located in the base station to use it when making scheduling decisions.…”

Stability with file arrivals and departures in multichannel cellular wireless networks

“…This observation has motivated recent interests in developing new scheduling algorithms for wireless networks with flow-level dynamics, and throughput-optimal scheduling algorithms for single-channel networks with flowlevel dynamics have been developed in [1,4,14,17,22].…”