Linear Precoding Bounds for Wyner-Type Cellular Networks With Limited Base-Station Cooperation and Dynamic Clustering

This work studies the achievable average data rate in the downlink of a Wyner-type linear cellular network with limited cooperation and linear precoders. We derive upper and lower bounds on the achievable average data rate subject to an average power constraint. The bounds show that even a small number of cooperating base stations applying suboptimal (linear) precoding can approach optimal performance, and significantly increase the average user data rates as compared to those achievable with noncooperating cellular systems. We highlight the importance of dynamic clustering and of base station (BS) silencing. Dynamic clustering allows the system to build clusters that benefit some users at the expense of others, and then shifts the priority between the users. Such a scheme results in higher average rates for all users, while meeting some level of fairness. BS silencing is shown to be important for limited cooperation networks in the high SNR regime. Silencing is a simple and efficient way to reduce the interference (and hence increase the user rates) without increasing the number of cooperating BSs. The theoretical analysis is accompanied by several numerical examples in selected scenarios.