IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 57, NO. 3, MAY2008 1789 A Novel Approach for Joint Radio Resource Management Based on Fuzzy Neural Methodology Lorenza Giupponi, Member, IEEE, Ramon Agustí, Member, IEEE, Jordi Pérez-Romero, Member, IEEE, and Oriol Sallent Roig Abstract—In this paper, an innovative mechanism to perform joint radio resource management (JRRM) in the context of hetero- geneous radio access networks is introduced. In particular, a fuzzy neural algorithm that is able to ensure certain quality-of-service (QoS) constraints in a multicell scenario deployment with three different radio access technologies (RATs), namely, the wireless local area network (WLAN), the universal mobile telecommunica- tion system (UMTS), and the global system for mobile communi- cations (GSM)/Enhanced Data rates for GSM Evolution (EDGE) radio access network (GERAN), is discussed. The proposed fuzzy neural JRRM algorithm is able to jointly manage the common available radio resources operating in two steps. The first step selects a suitable combination of cells built around the three avail- able RATs, while the second step chooses the most appropriate RAT to which a user should be attached. A proper granted bit rate is also selected for each user in the second step. Different implementations are presented and compared, showing that the envisaged fuzzy neural methodology framework, which is able to cope with the complexities and uncertainties of heterogeneous scenarios, could be a promising choice. Furthermore, simulation results show that the reinforcement learning mechanisms intro- duced in the proposed JRRM methodology allow guaranteeing the QoS requirement in terms of the so-called user dissatisfaction probability in the presence of different traffic loads and under different dynamic situations. Also, the proposed framework is able to take into consideration different operator policies as well as different subjective criteria by means of a multiple decision- making mechanism, such as balancing the traffic among the RATs or giving more priority to the selection of one RAT in front of another one. Index Terms—Beyond third-generation (3G) networks, fuzzy neural controllers, joint radio resource management (JRRM), radio access technology (RAT) selection. I. I NTRODUCTION W IRELESS mobile digital communication systems have been releasing services to the mass market for more than a decade, first focusing on voice service and, more re- cently, on a variety of data services. In this context, the problem faced by a network operator is to offer a system where the net- work usage is maximized for a given set of quality-of-service (QoS) requirements. In the traditional approach to solving this problem, two aspects can be clearly distinguished: network Manuscript received May 2, 2005; revised February 28, 2006, November 11, 2006, and July 27, 2007. This work has been performed within the framework of the EU funded project E2R. This work was supported in part by the Spanish Research Council under COGNOS Grant TEC2007-60985. The review of this paper was coordinated by Prof. S.-L. Kim. The authors are with Universitat Politècnica de Catalunya (UPC), Barcelona, Spain (e-mail: lorenza.giupponi@tsc.upc.edu; ramon@tsc.upc.edu; jorperez@tsc.upc.edu; sallent@tsc.upc.edu). Digital Object Identifier 10.1109/TVT.2007.907012 planning (i.e., the design of the fixed network infrastructure in terms of the number of cell sites, cell site location, number and architecture of concentration nodes, etc.) and radio resource management (RRM) (i.e., for a given network deployment, the way radio resources are dynamically managed in order to meet the instantaneous demand of the users moving around the network). In the framework of second-generation (2G) time-division multiple access (TDMA)-based mobile systems, e.g., global system for mobile communications (GSM), network planning is key. For a given network configuration, there is an almost constant value for the maximum capacity, and radio resource allocation actions in the short-term scale have a limited impact. On the contrary, in the framework of third-generation (3G) mobile systems, the situation is significantly different as long as code division multiple access (CDMA) becomes the dom- inant technology. The reasons are twofold. First, in CDMA- based systems, there is no constant value for the maximum available capacity since it is tightly coupled with the amount of interference in the air interface. Second, the multiservice scenario drops for some services the constant delay requirement and, consequently, opens the ability to exploit RRM functions to guarantee a certain target QoS, to maintain the planned coverage area, and to offer a high capacity while using the radio resources in an efficient way [1]. In turn, the perspective of Beyond 3G systems is that of heterogeneous networks, where the multiplicity of access tech- nologies as well as the diversity of terminals with reconfig- urability capabilities will be key in order to allow users on the move to enjoy seamless wireless services, irrespective of geographical location, speed, and time of day [2]. In this sce- nario, joint resource radio management (JRRM) is the identified process to manage dynamically and coordinately the allocation and deallocation of radio resources (e.g., time slots, codes, frequency carriers, etc.) between different radio access tech- nologies (RATs) for the spectrum bands allocated to each of these systems. With JRRM, a more efficient usage of the radio resources will follow. Some approaches to the JRRM problem are available in the open literature, and most of them are focused on functional and architectural behaviors. For example, [3] presents an Internet protocol (IP)-based end-to-end architecture involving different network domains where JRRM becomes a key element. In turn, [4] presents an interesting framework for the provision of JRRM algorithms to deal with the high degree of complex- ity associated with heterogeneous network scenarios. Another interesting contribution to JRRM can be encountered in [5], 0018-9545/$25.00 © 2008 IEEE