Study and development of advanced meta-heuristic approaches for resource allocation in intelligent communication systems

Abstract

The starting point of this dissertation lies on the concept of intelligent radio systems, which brings together all those communication systems capable of adapting their operations - via the optimized management of their available radio resources - to both the status of the physical medium through which the information is sent and possible restrictions imposed by the user, operator, or application at hand. The reasons for the implementation and deployment of this kind of systems are manyfold and diverse: as to mention, the mobility of the devices incorporating this adaptive functionality (ranging from people to vehicles of any kind) lead to amplitude fluctuations and eventually, deep fades in the signal power received at the destination front-end, with particular emphasis on wireless transmission mediums. Furthermore, the ecosystem of applications using wireless communication systems require an increasingly demanding quality of service by the equally growing size of the underlying networks supporting multiple and concurrent multimedia communications. Both factors determine significantly the allocation of radio resources (such as frequency, time or power) between users and applications, so they must be taken into consideration when optimally managing such resources. Although the advent of the so-called transmission adaptability dates back to the early nineties, the interest in intelligent radio systems has gained momentum in the last decade, mainly due to the crystallization of new technological trends in regulatory radio spectrum policies. Many studies have indeed shown that in recent years the inefficiency of allocation and resource exploitation strategies traditionally imposed throughout the World: the exclusive allocation and operation of certain “licensed” frequency bands by operators, and the arbitrary use of “unlicensed” spectral bands by operators and private users. To date the coexistence of both policies in the international regulatory framework has sown a fertile substrate for the rapid growth and development of wireless communication systems at both technical and application levels. However, the release of the so-called “digital dividend” (i.e. the UHF band between 790 and 862 MHz) has revealed numerous shortcomings of this regulatory framework, which call for the adoption of alternative mixed regulatory schemes. In addition, several measurement campaigns carried out in different countries (e.g. USA, United Kingdom and Finland) have unanimously concluded that although the licensed spectrum is considered a scarce and valued resource, a large portion is not utilized or when used, is done in short periods of intense use or bursts. In this context, cognitive radio stands emerges as a radio communication paradigm that can provide a flexible and practical technological response to the need for a more flexible and dynamic access to radio spectrum. Conceived as the most technologically advanced form of intelligent communication systems, cognitive radio devices are capable of monitoring the status of different spectral bands, determining their occupancy level, and opportunistically and autonomously adapting their radio parameters for the best use of the spectrum band at hand. The adaptability of radio resources and the development of algorithms for their efficient management lie consequently at the heart of current and future radio communication standards. This being exposed, this dissertation aims at shedding light on the applicability of novel meta-heuristic techniques for the optimization of the allocation of resources in intelligent communication systems. Inspired by the observation of physical phenomena and social behaviors in the Nature, these techniques are known to efficiently balance the tradeoff between computational complexity and the quality of the generated solutions, which is deemed of utmost relevance in view of the growing scales of communication networks. Specifically, three are the scenarios where meta-heuristics are exemplified as efficient algorithmic means for the management of radio resources: • The first scenario focuses on radio communication downlinks where a base station transmits information to several distant user through a single frequencywide orthogonally multiplexed transmit signal. The establishment of limits on the perceived transmission rate for each user and the insertion of a second base station as an unlicensed transmitter on the band in question yields a constrained optimization problem, which is efficiently solved by the novel meta-heuristic Harmony Search algorithm. This contribution also includes an innovative greedy repair procedure that accounts for the fulfillment of the produced solutions with respect to the imposed transmission rates by user. • The second stage addresses a dynamic channel allocation problem in cognitive radio networks, where Harmony Search is again employed for the allocation of frequency channels in the links between nodes under an error rate minimization criteria. The novelty of this contribution lies not only in the adaptation of the Harmony Search optimization technique itself to this particular scenario, but also in the tailored design of this algorithm for distributed network topologies, in an attempt at balancing the localized complexity featured by their centralized counterpart. • The third scenario goes beyond meta-heuristics to span the contributions of the Thesis towards another branch in the field of Soft Computing: fuzzy logic. Specifically, this contribution proposes a novel selection scheme for spectrum sensing methods according to a set of heuristic rules that can be made by experts or learnt from previous actions and their results. Fuzzy logic eases the joint consideration of several parameters of relevance to the selection process (e.g. detection probability or the available computation time), hence overriding the necessity for defining a precise mathematical model for the process itself. In summary, this dissertation takes a technical step towards the adoption of metaheuristic techniques – and by extension, Soft Computing methods – for the management of radio resources in intelligent communication systems.