1st International ICST Workshop on Multimode Wireless Access Networks

Research Article

RAT Selection Optimization in Heterogeneous Wireless Networks

Download
469 downloads
  • @INPROCEEDINGS{10.1007/978-3-642-30376-0_33,
        author={Angelos Rouskas and Pavlos Kosmides and Anastassios Kikilis and Miltiades Anagnostou},
        title={RAT Selection Optimization in Heterogeneous Wireless Networks},
        proceedings={1st International ICST Workshop on Multimode Wireless Access Networks},
        proceedings_a={MOWAN},
        year={2012},
        month={10},
        keywords={Next Generation Wireless Networks network selection optimization Branch and Bound},
        doi={10.1007/978-3-642-30376-0_33}
    }
    
  • Angelos Rouskas
    Pavlos Kosmides
    Anastassios Kikilis
    Miltiades Anagnostou
    Year: 2012
    RAT Selection Optimization in Heterogeneous Wireless Networks
    MOWAN
    Springer
    DOI: 10.1007/978-3-642-30376-0_33
Angelos Rouskas1,*, Pavlos Kosmides2, Anastassios Kikilis3,*, Miltiades Anagnostou2,*
  • 1: University of Piraeus
  • 2: National Technical University of Athens
  • 3: University of the Aegean
*Contact email: arouskas@unipi.gr, akikilis@aegean.gr, miltos@central.ntua.gr

Abstract

While wireless access networks are rapidly evolving, constantly increasing both in coverage and offered bandwidth, the vision for Next Generation Wireless Networks (NGWNs) encompasses a core network incorporating various Radio Access Technologies (RATs) in a unified and seamless manner. In such an environment, providers with multi-RAT technologies will aim at the maximization of the satisfaction of their subscribers, while attempting to avoid overloading their subsystems. In this paper we deal with the network selection problem in a multi-RAT environment where users are equipped with multimode terminals. We introduce a utility-based optimization function and formulate the problem of allocating user terminals to RATs as an optimization problem under demand and capacity constraints. This problem is recognized as NP-hard and we propose an optimal Branch and Bound (BB) algorithm, as well as a greedy heuristic which exploits a metric that measures the utility gained versus the resource spent for each allocation. BB manages to significantly reduce the search procedure, while greedy produces optimal allocation results similar to BB but with very low computational cost.