The operational spectral efficiency of wireless networks is typically far lower than the spectral efficiency of underlying physical layers. This reduced spectral efficiency is primarily due to protocol overhead, which consists of transmissions that do not depend on information bits, in both physical and network layers. However, none of our analytical models allow a coherent and complete accounting for protocol overheads in the system. Using the viewpoint that network protocols manage time-varying unknowns, this research is developing fundamentally new methods to account for all resource consumption in wireless networks.
The key innovation is the use of a two-way channel formulation to account for all resource usage, whether it is for feedback, state estimation or transmission of data. In the two-way formulation, each node has data for other nodes in the system, and hence all nodes are both receivers and transmitters. With the two-way structure at every node, data-independent transmissions are jointly coded with the data-dependent transmissions to allow a coding-theoretic analysis of finite, noisy side-information transmission. The investigators study both fading channels and bursty sources, and derive new performance bounds and protocol classes. The new protocols classes can then be used to derive practical protocols which dramatically reduce overhead in real wireless systems.
