ETD

• 4G
• AMC
• bit loading
• goodput
• greedy
• next generation
• OFDM
• optimization
• wireless

Technologies as MIMO and BIC-OFDM have proved attractive performance in harsh multipath scenarios, making them suitable candidates for satisfying the ambitious goals of next generation wireless communication systems. Such techniques require a great deal of parameters to be optimized, for providing the best performance in every possible scenario.
The purpose of this thesis is to investigate the performance that a wise bit distribution among the subcarriers of an OFDM system-or "bit loading"-making use of some CSI, could provide, relevant to that achievable by a system which employs a fixed modulation scheme. In doing this, we try to maximize the goodput, which represents the average number of bits belonging to error-free data received by unit of time.
As background, we review an heuristic technique called "greedy algorithms", whose key feature is that of trying to arrive at a global optimal solution of a problem, through a series of locally optimal (greedy) choices of many subproblems. So we make use of this methodology for elaborating some iterative sub-optimal bit loading procedures, and by means of simulation tools, we show that their contribution to the raise of the goodput performance is remarkable.
The elaborated methods are based on a PER estimation made with the Link Quality Metric "kESM", which has been proved to be a reliable link performance predictor, for multicarrier systems employing a variety of modulation schemes.