Buffer-Aided Relaying with Adaptive Link Selection

Abstract

In this paper, we consider a simple network consisting of a source, a half-duplex decode-and-forward relay, and a destination. We propose a new relaying protocol employing adaptive link selection, i.e., in any given time slot, based on the channel state information of the source-relay and the relay-destination link a decision is made whether the source or the relay transmits. In order to avoid data loss at the relay, adaptive link selection requires the relay to be equipped with a buffer such that data can be queued until the relay-destination link is selected for transmission. We study both delay constrained and delay unconstrained transmission. For the delay unconstrained case, we characterize the optimal link selection policy, derive the corresponding throughput, and develop an optimal power allocation scheme. For the delay constrained case, we propose to starve the buffer of the relay by choosing the decision threshold of the link selection policy smaller than the optimal one and derive a corresponding upper bound on the average delay. Furthermore, we propose a modified link selection protocol which avoids buffer overflow by limiting the queue size. Our analytical and numerical results show that buffer-aided relaying with adaptive link selection achieves significant throughput gains compared to conventional relaying protocols with and without buffers where the relay employs a fixed schedule for reception and transmission.

Radio Resource Allocation Over Fading Channels Under Statistical Delay Constraints

“…In general, the power allocation solution swings between the two policies. In addition, we show that, under very loose delay constraints, the allocation solutions converge to the solutions derived in [46] under unconstrained delay assumption. Illustrative results show that the proposed adaptive relaying outperforms fixed relaying under sufficiently loose delay constraints.…”

“…In [46], the authors consider the adaptive link selection relaying problem described as follows. Let φ[t] ∈ {0, 1}, ∀t denote a binary variable for frame t where we set φ[t] = 1 if the R-D link is active and φ[t] = 0 if the S-R link is active.…”

On Buffer-Aided Relaying for the Two-Hop Full Duplex Relay Channel with Self-Interference

“…This is because, when condition (13) holds, the number of time slots for which R RD (i) = min{Q(i − 1), C RD (i)} = Q(i − 1) occurs is negligible compared to the number of time slots for which R RD (i) = min{Q(i − 1), C RD (i)} = C RD (i) occurs when N → ∞, see [6]. In other words, when condition (13) holds, we can consider that the buffer at the relay has enough information almost always.…”

“…Specifically, [5] proposed a conventional decode-and-forward (DF) relaying scheme, where the HD relay switches between reception and transmission in a prefixed manner. On the other hand, [6] proposed a buffer-aided relaying scheme where, in each time slot, the HD relay selects to either receive or transmit based on the qualities of the receiving and transmitting channels. As a result, the rate achieved by the scheme in [6] is larger than the rate achieved by the scheme in [5], showing that buffers improve the performance of HD relays.…”

“…On the other hand, [6] proposed a buffer-aided relaying scheme where, in each time slot, the HD relay selects to either receive or transmit based on the qualities of the receiving and transmitting channels. As a result, the rate achieved by the scheme in [6] is larger than the rate achieved by the scheme in [5], showing that buffers improve the performance of HD relays. The capacity of the two-hop FD relay channel with an idealized FD relay without SI was derived in [1] and [5] for the cases with and without fading, respectively.…”