Network-assisted resource management for wireless data networks

We propose a framework for network-assisted radio resource management in wi reless data networks, This type of radio resource management techniques off er implementation and capacity benefits compared to conventional, interfere nce-measurement based, dynamic channel assignment (DCA) algorithms. The bas ic idea is to use interbase signaling to shift most of the burden of the re source allocation from the air interface to the backbone infrastructure. By exchanging channel assignment as well as other relevant information in rea l time through the backbone network, each base can calculate the impact of a resource assignment on the system. As a result, rapid interference measur ements, which are typically needed to implement DCA schemes, are replaced b y a limited amount of path loss measurements and the computation of interfe rence conditions by the base stations. This significantly reduces the measu rement and over-the-air signaling requirements, and can also provide an opp ortunity for a better optimization of the system performance. We focus on two specific algorithms: network-assisted least-interference-ba sed dynamic packet assignment (NA-LI-DPA) and network-assisted dynamic pack et assignment with throughput optimization (NA-DPA), NA-LI-DPA closely rese mbles a least-interference-based dynamic channel assignment algorithm, and NA-DPA attempts to further improve the overall system throughput, The algor ithms, as defined, are appropriate for a best-effort data service, where th e primary goal is to provide a higher throughput. However, it will be clear from the discussion that it is also feasible to alter the algorithms to op timize performance metrics other than throughput, e,g,, to ensure a certain quality of service, We show through simulation that, for a system like enh anced general packet radio service (EGPRS) system, NA-DPA can provide a thr oughput that is 50% higher than random packet assignment, and 25% higher th an that obtained by conventional DCA algorithms.