A study of cross layer trade-offs in QOS and security of wireless networks

This dissertation research is aimed at understanding the tradeoffs among different performance metrics of wireless networks namely, throughput, complexity, energy efficiency, Quality of Service (QoS), robustness, and security, and to formulate a class of algorithms to achieve the optimal tradeoffs. In this pursuit, the network is looked at as a single entity without emphasis on the layered architecture. Some of the techniques and algorithms developed in this dissertation are applicable to infrastructure-based as well as infrastructure-free (ad-hoc mode) networks. The proposed schemes are examined in the contexts of current wireless network standards. The layers involved it each case are identified, and the guidelines for cross-layer implementations are established. The performance evaluations are carried out with analysis and simulation.; First chapter of the dissertation provides an introduction to the work carried out in this dissertation. Chapter two addresses the complexity-throughput tradeoff in channel adaptive rate assignment and user scheduling for down-link in a multi user scenario. We propose the use of stochastic adaptation techniques namely discrete pursuit learning automata as a way to reduce the complexity and feedback requirements in achieving the optimal throughput. In chapter three, we examine the tradeoff between data encryption for privacy (confidentiality) and data throughput in a wireless link. Tradeoff optimal methods are proposed and analyzed under varying level of channel knowledge namely, (i) the exact channel states are known for a finite future (ii) the current channel state is known and a Markovian model of the channel is available, and (iii) only the average signal to noise ratio (SNR) of the channel and the probability distribution of the channel SNR are known.; In chapter four, we study the tradeoff in link scheduling for network throughput and connectivity to achieve reliable routing. While scheduling for throughput implies minimum redundancy in connectivity, reliability in routing is measured by the number of distinct alternative paths. We analyze the complexity involved in optimum scheduling of links to maximize the capacity and minimize the delay in providing transmission opportunity to every link in the network with a given connectivity goal. We show that although the scheduling problem with different constraint sets are NP-complete problems in general, a class of scheduling problems which is typical of the practical scenarios can be solved in polynomial time. Further in this chapter, we formulate several optimal and greedy algorithms with polynomial order complexity that achieve close to optimal throughput and are suitable to achieve the optimization goals in different scenarios.; A summary of the key results from this research is presented in chapter five.