The fading three-hop half-duplex relay network consists of a source, two half-duplex relays, and a destination connected in series where links are present only between adjacent nodes. We assume that the links are impaired by time-continuous fading and additive white Gaussian noise. For this network, we design new protocols based on buffer-aided relaying and derive their achievable average rates. We first develop a buffer-aided protocol which maximizes the average rate, but, as a side effect, introduces unbounded delay. Therefore, we also design a buffer-aided protocol which constrains the average delay, but at the expense of decrease of rate. Our numerical results show that the maximum average rate achieved with the developed buffer-aided protocol is larger than that of existing protocols for the considered network. Moreover, given a sufficiently large permissible average delay, the average rate achieved with the buffer-aided protocol with a delay constraint approaches the maximum average rate achieved without a delay constraint.
I. INTRODUCTIONFuture generation communication networks are expected to include some form of cooperative, relay-based communication. Employing relays in a network increases coverage and/or throughput and/or reliability. This phenomenon was first observed in [1], [2] for the simple three-node relay network comprised of a source, a full-duplex relay, and a destination. As the distance between the source and the destination increases, more than one relay have to be employed between the source and the destination in order to combat the large path loss caused by the long distance. A suitable network architecture that can be employed for such a scenario is the multi-hop relay network. The multi-hop relay network is comprised of a source, multiple relays connected in series, and a destination, where links are present only between adjacent nodes. As a result of this practical application scenario, multi-hop relaying has gained considerable research interest, see [3]- [9].The relays in the multi-hop relay network can be either fullduplex or half-duplex. The capacity of the full-duplex multi-hop relay network is known, and can be derived using the degraded relay channel model presented in [2]. However, due to selfinterference limitations, it is difficult to realize an ideal fullduplex relay in practice. As a result, half-duplex relaying is preferred for practical cooperative networks. However, the capacity for the multi-hop half-duplex relay network with additive white Gaussian noise (AWGN) is still unknown. To the best of our knowledge, the largest reported achievable rate for the AWGN half-duplex multi-hop relay network was given in [9]. This rate is achieved when the relays receive in one time slot and forward the received information in the following time slot. However, in fading environments, using relays which switch successively between reception and transmission degrades the achievable rate, as shown in [10] and [11] for the one-way and two-way three node relay channels, respectively. In [...
