Throughput Maximization for Delay-Sensitive Random Access Communication
share
We consider a single cell uplink in which many devices randomly transmit a data payload to the base station. Given a fixed latency constraint per device, we propose a time and frequency slotted random access scheme with retransmissions, which when necessary, are Chase combined at the receiver. We analyze the proposed setting at constant SNR. We characterize the scaling of random access throughput versus the latency, by optimizing the number of retransmissions and the number of frequency bins. For infinite block length (IBL), we conclude that at low SNR devices should completely share the time and frequency resources. For high SNR, however, the number of frequency bins should scale with altered load, and the slot duration for each retransmission is determined by the outage tolerance. Since infinite packet sizes are not possible, we extend our model to the finite block length (FBL) regime and characterize the gap versus the IBL regime. We also provide some new results for FBL capacity to bound the probability of outage. The proposed random access model gives an upper bound for the total uplink traffic that can be successfully decoded for a single receive antenna given the latency constraint, and provides insights for 5G cellular system design.
READ FULL TEXT