Optimisation of base station location in broadband power line communications access networks. (1st October 2010)

• Record Type: Journal Article
• Authors: Haidine, Abdelfatteh; Lehnert, Ralf
• Is Part Of: International journal of communication networks and distributed systems. Volume 5:Number 4(2010)
• Page Start: 347
• Page End: 374
• Publication Date: 2010-10-01
• Subjects: power line communications -- broadband networks -- B-PLC access networks -- network planning -- base station placement -- BSP -- multi-objective optimisation -- MOO -- evolutionary algorithms -- mathematical modelling -- network costs -- end-to-end delay -- metaheuristics -- evolutionary search; Computer networks -- Periodicals; Telecommunication systems -- Periodicals; Electronic data processing -- Distributed processing -- Periodicals; 004.6
• Languages: English
• ISSNs: 1754-3916

Abstract

In this paper, we investigate the planning and optimisation of the broadband power line communications (B-PLC) access networks. The B-PLC is a recent access technology that is considered as an alternative for the realisation of the broadband access networks, allowing bit rates up to 200 Mbps. This technique uses the wiring of the low voltage grid in order to offer to the users the telecommunications services, such as internet, voice-over-IP, video, etc. In this paper, the B-PLC planning process is subdivided into two parts, which are the generalised base station placement (GBSP) and the PLC channel allocation problem (P-CAP). This paper focuses on the GBSP that is formulated as multi-objective combinatorial optimisation. The analysis of this problem covers the problem description, the mathematical modelling of the different optimisation objectives (namely the network costs and the end-to-end delay – in uplink as well as in downlink) and its solution by means of metaheuristics. Two optimisation approaches are used to solve it: the classical approach based on single-objective optimisation and the modern approach based on multi-objective optimisation. Different algorithm variants are used, which are all based on the evolutionary search. The performances of these approaches are evaluated.