Code design based on metric-spectrum and applications
Extensive comparisons to metric bounds and to previously designed codes show the optimality or near-optimality of the new codes, designed for the fading and the additive white Gaussian noise channel (AWGN). It was also shown that linear codes can be optimal or at least meeting the metric bounds; one example is the systematic pilot-based code of rate k/(k + 1) which was proved to meet the lower bound on the maximum cross-correlation. Further, the method was generalized such that good codes for arbitrary channels can be designed given the corresponding metric or the pairwise error probability.
In synchronous multiple-access schemes it is common to use unitary block codes to transmit the multiple users? information, especially in the downlink. In this work we suggest the use of newly designed non-unitary block codes, resulting in increased throughput e?ciency, while the performance is shown not to be substantially sacri?ced. The non-unitary codes are again developed through suitable nested searches. In addition, new multiple-access codes are introduced that optimize certain criteria, such as the sum-rate capacity.
Finally, the introduction of the asymptotically optimum convolutional codes for a given constraint length, reduces dramatically the search size for good convolutional codes of a certain asymptotic performance, and the consequences to coded code-division multiple access (CDMA) system design are highlighted.
Advisor:Georghiades, Costas N.; Huang, Garng M.; Bettati, Riccardo; Miller, Scott L.
School:Texas A&M University
School Location:USA - Texas
Source Type:Master's Thesis
Keywords:code design block codes convolutional metric spectrum minimum distance maximum crosscorrelation awgn channel noncoherent fading coherent
Date of Publication:12/01/2004