Capacity per unit area of distributed antenna systems with centralized processing

We consider an extended wireless network with transmit and receive nodes distributed according to Bernoulli lattice processes in ID and 2D spaces. The received signals are jointly processed at a central unit. The channel is characterized by pathloss attenuation depending on distances between transmit and receive antennas. We introduce a new class of Euclidean random matrix (ERM) to characterize the distributed antenna system (DAS). By leveraging on a suitable decomposition of these ERMs, we propose an approximated analysis of their spectra and use it to provide an analytical approximation of the capacity per unit area of the DAS.