DocumentCode
991773
Title
Orthogonal Codes for Robust Low-Cost Communication
Author
Zhang, Wenyi ; Mitra, Urbashi
Author_Institution
Ming Hsieh Dept. of Electr. Eng., Univ. of Southern California, Los Angeles, CA
Volume
54
Issue
12
fYear
2008
Firstpage
5411
Lastpage
5426
Abstract
Orthogonal coding schemes, known to asymptotically achieve the capacity per unit cost (CPUC) for single-user ergodic memoryless channels with a zero-cost input symbol, are investigated for single-user compound memoryless channels, which exhibit uncertainties in their input-output statistical relationships. A minimax formulation is adopted to attain robustness. First, a class of achievable rates per unit cost (ARPUC) is derived, and its utility is demonstrated through several representative case studies. Second, when the uncertainty set of channel transition statistics satisfies a convexity property, optimization is performed over the class of ARPUC through utilizing results of minimax robustness. The resulting CPUC lower bound indicates the ultimate performance of the orthogonal coding scheme, and coincides with the CPUC under certain restrictive conditions. Finally, still under the convexity property, it is shown that the CPUC can generally be achieved, through utilizing a so-called mixed strategy in which an orthogonal code contains an appropriate composition of different nonzero-cost input symbols.
Keywords
channel capacity; minimax techniques; orthogonal codes; capacity per unit cost; channel transition statistics; input-output statistical relationships; minimax formulation; nonzero-cost input symbols; orthogonal codes; orthogonal coding schemes; robust low-cost communication; single-user ergodic memoryless channels; zero-cost input symbol; Channel capacity; Costs; Gaussian distribution; Memoryless systems; Minimax techniques; Random variables; Robustness; Statistics; Symmetric matrices; Uncertainty; Capacity per unit cost; channel uncertainty; compound channel; minimax robustness; minimax theorem; mixed strategy; orthogonal code;
fLanguage
English
Journal_Title
Information Theory, IEEE Transactions on
Publisher
ieee
ISSN
0018-9448
Type
jour
DOI
10.1109/TIT.2008.2006463
Filename
4675748
Link To Document