AG Kommunikationstheorie


Generalised multi-receiver radio network: QoS-feasibility, power-levels, power adjustment, and network simplification


We consider a generalised multi-receiver radio network under quality-of-service (QoS) constraints. We seek conditions under which the QoS targets are feasible, in the sense that there are power levels that can produce them. When the QoS targets are feasible we attempt to describe the corresponding power levels, either directly, by attempting to produce explicit formulae for them, or indirectly, by studying the convergence properties of a simple power-updating algorithm. Key to the analysis are the interference functions --- which take as input the a power vector and give the interference suffered by a transmitter at a given receiver --- and the QoS functions --- which take as input a vector of a transmitter's signal-to-interference ratios at each receiver and return a QoS value. This model includes as special cases many well-known schemes, such macro-diversity, multiple-connection reception and all others discussed by Yates (JSAC, 13(7):1341-1348, 1995).

The answers obtained depend on the specific assumptions made on the interference and QoS functions. In particular, if the key functions are monotonic and satisfy a mild homogeneity condition, a simple conservative feasibility condition for the QoS targets, and a power-vector that yields those targets are given in explicit form. The condition has the simple form, k_i <= q_i, where k_i is terminal i's desired QoS level, and q_i its quality of service when all power levels equal unity. If this condition is satisfied, the power levels P_i = k_i / q_i yield or exceed the desired QoS levels. Furthermore, the generalised multi-receiver network can be conservatively represented as an equivalent set of independent transmitter-receiver pairs, with q_i equal to the channel gain of the pair that represents i.

The additional assumption that the interference functions are sub-additive leads to a less conservative feasibility condition under which a simple, greedy power adjustment algorithm always converges to the unique power vector that produces the desired QoS targets. For the special case of macro-diversity --- the cellular structure is removed and all transmitters are jointly decoded by all receivers --- a simple macro-diversity feasibility condition that has some key advantages over that previously available emerges.

zurück zur Terminübersicht