A time-slotted-CDMA architecture and adaptive resource allocation method for connections with diverse QoS guarantees

  • Authors:

  • Samar Singh;Satish K. Tripathi

  • Affiliations:

  • Department of Computer Science and Engineering, La Trobe University, Melbourne, VIC 3086, Australia;Department of Computer Science, University of California, Riverside, CA

  • Venue:
  • Wireless Networks
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

We consider a time-slotted W-CDMA system for mobile stations which are connected to the wired internet. We first present an architecture for such a system that is based on a request-permission protocol incorporating power control for "Best Effort" transmissions on the uplink. The requesting mobiles are permitted to transmit in the next time slot with a specified power according to a schedule computed by the Base Station. To devise this scheduling method, we formulate a globally optimizing integer program that maximizes the total weighted sum of all best-effort transmissions in the entire system, keeping in view the diverse target Bit Error Rates for each one. This problem is analysed and decomposed into sub-problems that can be solved locally by each Base Station. We devise two fast heuristics to solve the Base Station's sub-problem, so that the new schedule for each successive slot can be re-computed by each Base Station in a practical timeframe. We show that one heuristic is good enough to produce optimal solutions to the sub-problem in special cases. The method is further enhanced to take account of bandwidth and delay guarantees for other connections. It is also modified to ensure fairness for best-effort code channels suffering from persistent location-dependent errors. Finally, we show that a very similar approach can be used by the Base Station for scheduling on the downlink also, leading to a unified approach to scheduling in both directions. The efficacy of the uplink method is briefly demonstrated by simulations comparing the two variants with each other, and demonstrating that one achieves a consistently higher throughput than the other.