Dynamic RAT selection for multiple calls in heterogeneous wireless networks using group decision-making technique

Abstract

Existing radio access technology (RAT)-selection algorithms for heterogeneous wireless networks (HWNs) do not consider the problem of RAT selection for a group of calls from a multimode terminal (MT). Multimode terminals (MTs) for next generation wireless networks have the capability to support two or more classes of calls simultaneously. When a new call is initiated on an MT already having an ongoing call in an HWN, the current RAT may no longer be suitable for the two calls (incoming call and the existing call). Thus, a new RAT may be more suitable for the two calls. The problem of RAT selection for two or more calls from an MT in an HWN is a group decision problem. This paper addresses the problem of RAT selection for a group of calls from an MT in an HWN by using the modified TOPSIS group decision-making technique. The paper proposes a dynamic RAT-selection algorithm that selects the most suitable RAT for a single call or group of calls from an MT in an HWN. The algorithm considers users’ preferences for individual RATs, which vary with each class of calls, in making RAT selection decisions in an HWN. A user’s preference for each of the available RATs is specified by weights assigned by the user to RAT selection criteria for different classes of calls. Based on the assigned weights, the proposed algorithm aggregates individual calls’ weights specified by the user to make a RAT-selection decision for a group of calls. In order to reduce the frequency of vertical handover, the proposed algorithm uses RAT preference margin in making RAT selection decisions. RAT preference margin is a measure of the degree to which the newly preferred RAT is better than the current RAT. Performance of the proposed algorithm is evaluated through numerical simulations. Results are given to show the effectiveness of the proposed RAT-selection algorithm.

Introduction

Joint radio resource management (JRRM) has been proposed for efficient radio resource utilization and enhanced QoS provisioning in heterogeneous wireless networks (HWNs), and a number of JRRM algorithms have been developed for HWNs [1], [2], [3]. The radio access technology (RAT) selection algorithm is one of the JRRM algorithms. The purpose of a RAT selection algorithm is to select the most suitable RAT for an incoming call in an HWN. Fig. 1 illustrates the problem of RAT selection for a single call in HWNs.

A number of RAT-selection algorithms have been proposed in the literature and these algorithms can be classified as single-criterion or multi-criteria RAT-selection algorithms. Single-criterion RAT selection algorithms use a single criterion for RAT selection whereas multi-criteria RAT selection algorithms use two or more criteria for RAT selection. Multi-criteria RAT-selection algorithms have been shown to be more efficient than single-criterion RAT selection algorithm. Thus much effort has been concentrated on developing multi-criteria RAT selection algorithms for HWNs [1], [3], [4], [5], [6], [7].

In [1], Giupponi and Pérez-Romero proposed a JRRM algorithm based on fuzzy neural approach for selecting the most appropriate RAT for an incoming call in HWNs.

In [3], Alkhawlani and Hussein proposed an intelligent RAT-selection algorithm for next generation wireless networks. The proposed algorithm uses a combined parallel fuzzy logic control and multi-criteria decision-making technique to select the most appropriate RAT for an incoming call in HWNs.

In [4], Zhang proposed a fuzzy multiple attribute decision-making (MADM) RAT-selection algorithm that uses fuzzy logic to represent imprecise information of some RAT-selection criteria. The fuzzy MADM method operates in two steps. The first step is to convert the imprecise fuzzy variables to crisp numbers. The second step is to use classical MADM technique to determine the ranking order of the candidate networks. The highest-ranking RAT is then selected for the call.

In [5], Xavier et al. presented a Markovian approach for RAT selection in an HWN. They developed an analytical model for a RAT-selection algorithm in an HWN comprising GSM/EGDE and UMTS. The proposed algorithm selects just one RAT for each incoming call.

In [6], Guo et al. proposed a RAT selection algorithm that uses a fuzzy multiple objective decision-making technique to select the most suitable RAT for each incoming call in an HWN.

In [7], Wu and Sandrasegaran conducted a study of RAT selection algorithm in a heterogeneous UMTS-GSM network.

All the RAT-selection algorithms reviewed above were designed to select the most suitable RAT for just one incoming call in HWNs. None of the RAT-selection algorithms considered the problem of RAT selection for a group of calls (multiple calls) from a multimode terminal (MT) in HWNs.

Multimode terminals (MTs) for next generation wireless networks have the capability to support two or more classes of calls simultaneously. Thus in a multi-service HWN, a subscriber using a multimode terminal (MT) will be able to access multiple services (such as voice, web session, video streaming, etc.) simultaneously, through any of the available RATs. Therefore, a group-call RAT selection problem occurs when a single RAT is to be selected for multiple classes of calls initiated from a multimode terminal in an HWN or when multiple classes of calls from a multimode terminal are to be handed over from one RAT to another.

Fig. 2 illustrates the problem considered in this paper, where a RAT is to be selected for multiple calls from an MT in an HWN.

As shown in Fig. 2, a user having an MT can initiate multiple calls from the MT, and a RAT has to be selected for the multiple calls in the HWN based on some RAT selection criteria. Different classes of calls have different QoS requirements, and different RATs in HWNs usually have different capabilities in terms of available data rate (bit per second), battery power consumption, security level provided, delay, etc. Selecting the most appropriate RAT for multiple calls from the MT in an HWN is a group decision problem. Existing RAT selection algorithms were not designed to select a RAT for multiple calls in HWNs.

Therefore, this paper addresses the problem of RAT selection for multiple calls from an MT using the modified TOPSIS group decision-making technique [8], [9].

The following are some reasons why it may be necessary to select a single RAT for multiple calls from a multimode terminal in HWNs: (1) to reduce handoff complexity, (2) to reduce signaling overhead, (3) to reduce battery power consumption, and (4) to accommodate low-capability multimode terminals.

If multiples calls from an MT are admitted into different RATs in an HWN, coordination of handover procedure among the different RATs will be complicated, and incur excessive signaling overhead. Moreover, multiple RAT interfaces of the multimode mobile terminals will be activated, which may increase the overall battery power consumption of the multimode terminal. In addition, some multimode terminals can be connected to only one RAT at a time. If these low-capability multimode terminals are to support multiple services, group decision is inevitable. Thus, it is necessary to develop an algorithm that will select the most suitable RAT for a group of calls from an MT in HWNs.

The objective of this paper is to develop a dynamic RAT-selection algorithm for making group call RAT-selection decisions in HWNs. The main contributions of this paper are threefold. The first is the conceptualization of group-call RAT-selection problem in HWNs. The second contribution is the development of a dynamic RAT selection algorithm and application of the modified TOPSIS group decision technique to solve the problem of RAT selection for multiple calls in HWNs. The third contribution is the investigation of the effect of RAT preference margin on the frequency of vertical handoff in HWNs. To the best of our knowledge, this is the first paper considering the problem of dynamic RAT selection for a group of calls from a multimode terminal in HWNs, and it is the first attempt to solve the problem.

The rest of this paper is organized as follows. In Section 2, multi-criteria group decision-making techniques are briefly reviewed. In Section 3, the proposed RAT selection algorithm is described. In Section 4, the modified TOPSIS group decision-making technique is applied to solve the problem of RAT selection for multiple calls in HWNs. Performance evaluation and simulation results are presented in Section 5.