Distributed Demand-Aware Access Selection in Wireless Multi-Cell Data Networks

Regardless of advances in transmission technology, wireless broadband access will result in that fewer active users can be supported in a given cell. This may, due to lower levels of statistical multiplexing, yield in an unbalanced network. For such contexts load aware access point (AP) selection (load balancing) has been proposed as a means to increase network performance. This paper evaluates the downlink capacity (maximum number of users that can be admitted given a throughput requirement) of distributed load aware AP selection criteria for a well planned network, where the expected number of users per cell coincide. We propose a market based algorithm in which AP selection is aided by market mechanisms. In contrast to most of the existing research, we account for that interference levels generated in different cells depend on user assignment, and therefore varies as terminals perform handoff. Compared to single-frequency systems where users base their selection on the received signal strength our results show that the downlink capacity can be increased with around 25 percent by introducing a channel plan in combination with demand, or load, aware AP selection criteria. This gain is fairly insensitive to the throughput requirement and does not come at the expense of uplink performance. To benefit from demand metrics it is, however, of paramount importance that the entire system bandwidth is not reused in all cells.