Analysis and Optimization of HARQ for URLLC

Abstract

In this paper, we investigate the effectiveness of the hybrid automatic repeat request (HARQ) technique in providing high-reliability and low-latency in the finite blocklength (FBL) regime in a single user uplink scenario. We characterize the packet error rate (PER), throughput, and delay performance of chase combining HARQ (CC-HARQ) and incremental redundancy HARQ (IR-HARQ) in AWGN and Rayleigh fading channel with m retransmissions. Furthermore, we consider a quasi-static fading channel model, which is more accurate than the oversimplified i.i.d. block fading or same channel assumption over consecutive packets. We use finite state Markov model under the FBL regime to model correlative fading. Numerical results present interesting insight into the reliability-latency trade-off of HARQ. Furthermore, we formulate an optimization problem to maximize the throughput performance of IR-HARQ by reducing excessive retransmission overhead for a target packet error performance under different SNRs, Doppler frequencies, and rate regimes.

HARQ Optimization for Real-Time Remote Estimation in Wireless Networked Control

“…We drop the time index k because the state transition function is time-homogeneous. Also in (15), ǫ (m) ir is the concise notation corresponding to the error probability of IR-HARQ in finite block length at the m-th transmission given in (5) as…”

“…The equivalent MDP problem of [11,Eq. (25)] can be obtained by fixing maximum retransmissions to 1 and τ = 1 in (15). Therefore, the IR-HARQ based approach adopted in [11] is a special case of the above proposed design when A {0, 1} and τ 1 = 1.…”

“…The standard CC-HARQ based optimal policy can be obtained using the MDP formulation for IR-HARQ by modifying the transition function (15) with the error probability ǫcc of CC-HARQ given in (4) and keeping τ = 1. In standard CC-HARQ, the full packet is repeated in response to retransmission requests, and the receiver increases the reliability using MRC.…”

“…Therefore, in the latest specifications of 5G for low-latency communication, only 1 or 2 retransmissions is supported [17]. Both IR-HARQ and CC-HARQ are studied in the finite blocklength (FBL) regime for their effectiveness in low-latency communication [15], [18]. HARQ is also optimized to meet strict packet-level latency requirements in Rayleigh fading channel [19], [20].…”

“…In the standard ARQ protocol, the sensor always sends the latest available update, and if the packet fails, the receiver discards that packet altogether. However, hybrid automatic repeat request (HARQ) utilizes forward error correction coding to facilitate combining the failing packet with its retransmissions to increase the reliability of an estimate [14], [15]. Given some knowledge of the dynamic process, the policy of sending new or retransmitting old updates is used to minimize the average MSE performance of such systems [11].…”