Elsevier

Physical Communication

Volume 18, Part 2, March 2016, Pages 125-139
Physical Communication

Full length article
Radio access technology selection in heterogeneous networks

https://doi.org/10.1016/j.phycom.2015.10.004Get rights and content

Abstract

The migration of wireless networking towards the 5G era is distinguished by the proliferation of various Radio Access Technologies (RAT). As no existing technology can be surrogated by another one, the coexistence of today wireless networks is the best solution at hand when dealing with the incessantly growing user demand for bandwidth. Hence, in this heterogeneous environment, users will be able to utilize services through diverse RATs. RAT selection is crucial and must be designed astutely to avoid resource wastage. In this paper, we consider the downlink of a heterogeneous network with two broadband RATs: a primary RAT such as LTE, and a secondary RAT such as WiFi. We start by formulating a centralized approach for the RAT selection as an optimization problem. Then, two distributed approaches are proposed for adequate RAT selection: first, we put forward distributed heuristic algorithms based on the peak rate perceived by users from available RATs. Second, we devise a distributed RAT selection scheme portrayed as a non-cooperative game with a learning-based algorithm to reach the Nash Equilibriums of the RAT selection game. Extensive simulation results show that the proposed distributed algorithms give efficient results compared to the centralized optimal approach. The analysis of the simulation results enables to define pertinent use cases that delimit the scope of the proposed optimal centralized and distributed approaches.

Introduction

Every year, the demand in mobile broadband communications increases spectacularly. Practical solutions need to be proposed to face this imminent thousand-fold traffic augmentation. To address this challenge, ubiquitous radio access will be offered by forthcoming 5G heterogeneous network deployments. On the one hand, the advent of mmWave technology and carrier aggregation mechanisms is inevitable to support higher capacity  [1]. On the other hand, improved spectral efficiency and novel heterogeneous network deployments with astute resource sharing are vital to meet the predicted traffic demands for the next decade.

For increased efficiency, heterogeneous networks will be self-organized. Operators will profit from the abundance of diverse Radio Access Technologies (RATs) in the same operating area and devise advanced Radio Resource Management (RRM) schemes to take advantage of the available system resources. Hence, RATs will need to be integrated, with any combination of 3G, WiFi  [2], WiMAX  [3] and LTE  [4]. As highlighted by the seminal paper  [5], heterogeneous networks are undeniably presented as a major cornerstone of the upcoming 5G networks. The authors particularly focus on the challenge of integrating different RATs. Moreover, the Horizon 2020 [6] European framework programme for research and innovation identifies that future networks will need to become significantly more heterogeneous and use multiple RATs. This challenge is tackled particularly by the METIS project that lays down the foundations of 5G networks  [7]. In such a heterogeneous network, when a new or a handover session arrives, a decision must be astutely made as to which technology it should be associated with. This is known as RAT Selection. In such a context, a mobile user will be able to connect concurrently to different RATs by enabling device support for carrier aggregation.

The straightforward approach in apprehending the RAT selection issue is to formulate the problem as centralized optimization task whose objective is to maximize throughput or equivalently minimize delay. In order to derive the expression of the delay, we use an analytical model whose key feature lies in accounting for the effect of interference as well as for the physical layer and channel characteristics in an easy and straightforward manner. On the one hand, the model takes into consideration frequency planning and scheduling aspects; and on the other hand, it provides tractable formulas of the end user mean delay.

While optimization models give an insight into the upper bounds of achievable RAT selection gains, the implementation of these centralized optimal mechanisms are cost prohibitive in real systems. Indeed, RRM mechanisms studied in the state-of-the-art build upon markedly lower complexity distributed schemes. Consequently, the present work is threefold: the first part addresses the RAT selection issue as a centralized optimization problem. The second part proposes simple but cost effective and fully distributed heuristic algorithms. The third part resorts to non-cooperative game theory to put forward a distributed algorithm based on replication dynamics where each mobile user selfishly strives to improve its own performances.

Results are validated through extensive simulations in the practical setting of a geographical area covered by a global LTE network acting as the primary RAT overlapping with several local WiFi hotspots acting as the secondary RAT. This typically corresponds to a WiFi offloading scenario  [8]. We begin by examining static scenarios chosen randomly then assess the algorithms performances in a dynamic setting.

The paper is organized as follows. Related work is presented in Section  2. The system model is described in Section  3 with adopted models for network structure, traffic, perceived rates, and cost function. The optimal centralized RAT selection scheme is formulated in Section  4 as a non linear optimization problem. The heuristic distributed approach is explained in Section  6. Furthermore, the RAT selection policy performances are assessed in a realistic dynamic setting. The game theoretic distributed approach is explained in Section  7 and a realistic distributed algorithm based on replicator dynamics is explained in 7.1 to reach the Nash equilibriums of the RAT selection game. Finally, in Section  8, simulations were conducted in a dynamic setting to compare all formulated approaches. We conclude in Section  9.

Section snippets

Related work

The need to fully profit from the large number of currently available RATs is the main driver behind the growing relevance of heterogeneity for future 5G networks. The subject is not only a hot topic for the scientific community but also for the related standardization bodies that are duly specifying procedures to support the interoperability between heterogeneous networks. In fact, the IEEE 802.21 group has defined [9], [10] a framework to enable seamless handovers between RATs.

In the

Network model on the downlink

We consider the downlink of a heterogeneous network with two broadband RATs: a primary RAT such as LTE, and a secondary RAT serviced by WiFi. An LTE cell range is in the order of a few kilometers while the WiFi cell range spans from a few tens to a few hundreds of meters only  [26]. Hence, typically, an LTE cell will be covered with several WiFi antennas as in Fig. 1. We will use the term BS (Base Station) to designate the serving antenna in any RAT: a WiFi antenna of an LTE BS.

We assume that

The optimal RAT selection problem

In the present section, we formulate a centralized approach for the RAT selection as a global non-linear optimization and analyze its salient properties. Based on these properties, we examine different algorithmic solutions based on an exhaustive search and on a Mixed Integer Linear Program (MILP) reformulation. We finally introduce a novel enhanced search algorithm that enables to drastically reduce the computation complexity. We assume the existence of a central entity responsible of routing

Cell decomposition

In this section, we propose to discretize the coverage area into zones characterized by similar radio conditions, in particular similar peak rates, as illustrated in Fig. 2. The reason behind this decomposition is to make the computation tractable. In fact, the interference and the path-loss factor are different for every user depending on its position in the cell. In practical network dimensioning, it is not possible to use, for each user, exact values for these parameters but average values

Heuristic distributed approach

Although optimal, the centralized approach presented in the previous sections can be costly and resource consuming (even for the enhanced search algorithm). From a system design perspective, distributed mechanisms using pre-configured and simple resource allocation rules are quite appealing. Therefore, in Section  6.1, we propose two lightweight heuristic distributed algorithms that approximate the optimal solution. In Section  6.2, extensive simulations are conducted to compare the two

Game theoretic distributed approach

To fully assess the relevance of our distributed heuristics in Section  6, we will compare their performances against another distributed RAT selection scheme based on non-cooperative game theory that we proposed in  [37]. Game theory is well adapted to model the interactions between players competing selfishly for a common resource. Initially, the game consists for each end-user in allocating the traffic among the primary and secondary RATs in a way to minimize selfishly its own cost. However,

Global comparisons

In this final section, extensive simulations are conducted in a dynamic setting to compare the heuristic algorithms R and PR, the replicator dynamics algorithm and the optimal centralized approach O. The probability vectors are respectively [0 0 0 0 0 0.1 0.3 0.6] in LTE and [0.7 0.3 0 0 0 0 0 0] in WiFi. The arrival of users follows a Poisson distribution and each user leaves the system after downloading a file whose size follows an exponential distribution of mean 5 MBytes. It is worth

Conclusion

Undeniably, ubiquitous radio access remains the essential backbone for supporting the ever increasing demand for bandwidth. Operators will profit from the abundance of diverse air interfaces in the same operating area and put forward advanced RAT selection mechanisms. The stringent performance targets and the novel flat architecture of beyond 4G networks have triggered a new trend in RRM favoring lightweight distributed schemes to costly centralized schemes. Hence, users are intelligent

Kinda Khawam got her engineering degree from Ecole Supérieure des Ingénieurs de Beyrouth (ESIB) in 2002, the Master’s degree in computer networks from Telecom ParisTech (ENST), Paris, France, in 2003, and the Ph.D. from the same school in 2006. She was a post doctoral fellow researcher in France Telecom, Issy-Les-Moulineau, France in 2007. Actually, she is an associate professor and researcher at the University of Versailles in France. Her research interests include radio resource management,

References (40)

  • L. Pirmez et al.

    SUTIL—Network selection based on utility function and integer linear programming

    Comp. Netw.

    (2010)
  • NGMN Alliance, 5G White Paper, February,...
  • Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) specifications, ANSI/IEEE Std 802.11,...
  • IEEE Standard for Local and Metropolitan Area Networks Part 16: Air Interface for Fixed Broadband Wireless Access...
  • S. Sesia et al.

    LTE—The UMTS Long Term Evolution: From Theory to Practice

    (2009)
  • J.G. Andrews et al.

    What will 5G Be

    IEEE J. Sel. Areas Commun.

    (2014)
  • ADVA et al., Horizon 2020 Advanced 5G Network Infrastructure for Future Internet PPP Industry Proposal (Draft Version...
  • METIS D6.6, Final Report on the METIS 5G System Concept and Technology Roadmap, ICT-317669 METIS Deliverable 6.6,...
  • F. Rebecchi et al.

    Data offloading techniques in cellular networks: A survey

    IEEE Commun. Surv. Tutor.

    (2014)
  • IEEE Standard for Local and Metropolitan Area Networks - Part 21: Media Independent Handover,...
  • K. Taniuchi, et al. IEEE 802.21: Media Independent Handover: Features, Applicability, and Realization,...
  • D. Kumar, E. Altman, J.-M. Kelif, User-Network Association in a WLAN-UMTS Hybrid Cell: Global & Individual Optimality,...
  • E. Stevens-Navarro et al.

    An MDP-based vertical handoff decision algorithm for heterogeneous wireless networks

    IEEE Trans. Veh. Technol.

    (2008)
  • M. Ibrahim, K. Khawam, S. Tohme, Network-centric joint radio resource policy in heterogeneous wimax-umts networks for...
  • G.H. Carvalho, et al. A joint call admission control-based approach for initial RAT selection in HetNets, in:...
  • I. Tsompanidis, A. Zahran, C. Sreenan, Towards utility-based resource management in heterogeneous wireless networks,...
  • J. Choque, R. Aguero, E.-M. Hortiguela, L. Munoz, Optimum Selection of Access Networks within Heterogeneous Wireless...
  • A. Kumar et al.

    A probabilistic approach to modeling users’ network selection in the presence of heterogeneous wireless networks

    IEEE Trans. Veh. Technol.

    (2014)
  • O. Eng Hwee et al.

    On optimal network selection in a dynamic multi-RAT environment

    IEEE Commun. Lett.

    (2010)
  • D. Kumar et al.

    User-network Association in an 802.11 WLAN and 3G UMTS Hybrid Cell: Individual Optimality

  • Cited by (0)

    Kinda Khawam got her engineering degree from Ecole Supérieure des Ingénieurs de Beyrouth (ESIB) in 2002, the Master’s degree in computer networks from Telecom ParisTech (ENST), Paris, France, in 2003, and the Ph.D. from the same school in 2006. She was a post doctoral fellow researcher in France Telecom, Issy-Les-Moulineau, France in 2007. Actually, she is an associate professor and researcher at the University of Versailles in France. Her research interests include radio resource management, modeling and performance evaluation of mobile networks.

    Samer Lahoud received the Ph.D. degree in communication networks from Telecom Bretagne, Rennes. After his Ph.D. degree, he spent one year at Alcatel-Lucent Bell Labs Europe. In 2007, he joined the University of Rennes 1 as an assistant professor. His research activities at the IRISA laboratory in Rennes focus on routing and resource allocation algorithms for wired and wireless communication networks. He has co-authored more than 60 papers published in international journals and conference proceedings.

    Marc Ibrahim is an assistant professor at Saint Joseph University of Beirut in Lebanon and the director of CIMTI (Centre d’Informatique, de Modélisation et de Technologies de l’Information). He got his engineering degree from the Faculty of Engineering at Saint Joseph University in 2002, then his Master’s degree from the same faculty in 2004. In 2009, He received his Ph.D. in communication networks from the University of Versailles in France. His research activities orbit wireless networks with a particular focus on radio resource management and network selection as well as performance modeling and networks measurement.

    Mohamad Yassin obtained his Ph.D. degree in Communication Networks jointly from University of Rennes 1 in France and Saint Joseph University of Beirut in Lebanon in 2015. He obtained his Engineering Diploma in Telecommunications and Networks from Saint Joseph University of Beirut, in 2011 and a Research Master’s in Telecommunication Networks jointly from the Lebanese University and Saint Joseph University of Beirut in 2012. His main research interests include Software Defined Networks (SDN), Network Functions Virtualization (NFV), Inter-Cell Interference Coordination (ICIC), Radio Resource Management (RRM), wireless Heterogeneous Networks (HetNets) and system-level simulation of mobile networks.

    Steven Martin graduated from ESIEE engineering school in 2000 and received his Ph.D. degree from INRIA, France, in 2004. Since 2005, Steven MARTIN is working at Paris-Sud University. His research interests include quality of service, LTE networks, network coding, ad hoc networks and real-time scheduling. Head of research group ROCS - Networking and Optimization - at LRI (Computer Science Laboratory, CNRS/Paris-Sud University), Pr. Martin is involved in many research projects and had the lead of the activity “Personal safety in digital cities of the future” in EIT ICT Labs (the European Institute of Innovation and Technology) from 2010 to 2014. He is the author of a large number of papers published in leading conference proceedings and journals. He has served as TPC member for many international conferences in networking.

    Melhem El Helou received the engineering and master degrees in communications and networking from the Ecole Supérieure d’Ingénieurs de Beyrouth (ESIB), Faculty of Engineering, Saint Joseph University of Beirut, Beirut, Lebanon, in 2009 and 2010, respectively, and the Ph.D. degree in communication networks from IRISA Research Institute, University of Rennes 1, Rennes, France and Saint Joseph University of Beirut, in 2014. He joined ESIB in September 2013 where he is currently an Assistant Professor. His research interests include wireless networking, radio access optimization, radio and energy resource management, and quality of service.

    Farah Moety is currently a post-doctoral fellow researcher in Orange Labs, Issy-Les-Moulineaux, France. She obtained her Ph.D. in 2014 from University of Rennes 1, Rennes, France. She was a member of ATNET research team at IRISA Labs, Rennes, France. She obtained her Master 2 in Telecommunication Networks in 2011 jointly from Lebanese University (EDST) and Saint-Joseph University (ESIB) in Lebanon. She has done her master thesis at IRISA Labs, Rennes, France. She received her Telecommunication and Computer Engineering diploma from Lebanese University — Faculty of Engineering III in 2010, Beirut, Lebanon. Her research interests include green wireless access networks, radio resource allocation, heterogeneous networks, LTE networks, WLANs, network optimization, radio access technology selection.

    View full text