Low latency communications

Numerous applications demand communication schemes that minimize the transmission delay while achieving a given level of reliability. An extreme case is high-frequency trading whereby saving a fraction of millisecond over a route between Chicago and New York can be a game-changer. While such communications are often carried by fiber, microwave links can reduce transmission delays over large distances due to more direct routes and faster wave propagation. In order to bridge large distances, information is sent over a multihop relay network. Motivated by these applications, this papers present an information-theoretic approach to the design of optimal multihop microwave networks that minimizes end-to-end transmission delay. To characterize the delay introduced by coding, we derive error exponents achievable in multihop networks. We formulate and solve an optimization problem that determines optimal selection of amplify-and-forward and decode-and-forward relays. We present the optimal solution for several examples of networks. We prove that in high SNR the optimum transmission scheme is for all relays to perform amplify-and-forward. We then analyze the impact of deploying noisy feedback.