A Clustering Approach to Planning Base Station and Relay Station Locations in IEEE 802.16j Multi-Hop Relay Networks

Abstract

In this paper, a clustering approach to solve a network planning problem for 802.16j relay networks is considered. Our clustering approach consists of three basic steps: (1) divide the nodes into k distinct clusters, (2) solve the planning problem separately for each cluster, and (3) perform a final optimization to reduce issues arising at cluster boundaries. Simulation results show that our approach is more efficient than existing approaches: solutions of equivalent quality can be found in 40% of the time. Thus our technique can be used to solve larger problems with similar hardware, or similar size problems in less time.

Uplink Capacity of Multi-Class IEEE 802.16j Relay Networks with Adaptive Modulation and Coding

“…Reference [1] & [2] show that with appropriate deployment of RSs in an MMR network, significant performance gains can be obtained in terms of the user throughput as well as the cellular coverage compared to conventional single-hop cellular networks. The capacity gain of an 802.16j system is investigated in [4] via simulations.…”

Comparison of Radio Wave Propagation Models for Mobile Networks

“…Frequencies are allocated within the cluster such that they minimize interference with adjacent channels [6]. These considerations constitute an initial planning phase for the network.…”

“…In theory, radio models can be arbitrarily complex. Working with such models, it is important to find a suitable level of abstraction, i.e., capturing a small number of characteristics may result in a shorter computational time but poor results, and vice versa [6].…”

Min-cut based partitioning for urban LTE cell site planning

“…In [9] and [10], Yu et al apply a k-means clustering method to partition instances for planning 802.16j base station and relay station locations into smaller sub-problems. Since the considered path loss models are semi-empirical, computation of the distance measure to new points, i.e., cluster centroids, is simple.…”