Transport capacity and spectral efficiency of large wireless CDMA ad hoc networks

Summary form only given. We study transport capacity and spectral efficiency of large wireless CDMA ad hoc networks. The former is defined as transported bit-meters per symbol period per node and the latter is bit-meters per Hz per second per volume meter under the joint optimization of coding and routing. The network consists of infinitely many nodes uniformly distributed in an n-dimensional space. Node density and spreading gain tend to infinity. The MMSE, decorrelator, or MF receiver is employed at a receive node. It is shown that the limit network at any time is composed of infinitely dense memoryless Gaussian channels each connecting a pair of send and receive nodes. Then the transport capacity, spectral efficiency and average number of hops are obtained. The network presents a "scaling law" in that the transport capacity vanishes as node density tends to infinity at a rate higher than processing gain. The "scaling law" disappears if processing gain increases as fast as node density does, which is in sharp contrast to the Gupta-Kumar network where a scaling law cannot be overcome. It is also shown that a supersaturated network where the traffic intensity defined as the number of active nodes per Hz per second per volume meter tends to infinity achieves the highest spectral efficiency but a vanishing transport capacity. On the other hand, as the traffic intensity vanishes, the spectral efficiency vanishes and the transport capacity achieves the highest. For a low-powered sensor network to achieve high spectral efficiency under the constraint of fixed total transmission power per volume meter the node density shall be increased to the maximum and the transmission power of each node be decreased to the minimum.