Distributed algorithms for optimal rate-reliability tradeoff in networks

The current framework of network utility maximization for distributed rate allocation assumes fixed channel code rates. However, by adapting the physical layer channel coding, different rate-reliability tradeoffs can be achieved on each link and for each end user. Consider a network where each user has a utility function that depends on both signal quality and data rate, and each link may provide a ´fatter´ (´thinner´) information ´pipe´ by allowing a higher (lower) decoding error probability. We propose two distributed, pricing-based algorithms to attain optimal rate-reliability tradeoff, with an interpretation that each user provides its willingness to pay for reliability to the network and the network feeds back congestion prices to users. The proposed algorithms converge to a tradeoff point between rate and reliability, which is proved to be globally optimal for codes with sufficiently large codeword lengths and user utilities with sufficiently negative curvatures