TY  - JOUR
T1  - High-Quality Link Adaptation Mechanism Based on LSTM in 802.11p V2X Communication
AU  - Kwon, Sang-Won 
AU  - Kim, Su-Hyun 
AU  - Lim, Jea-Han 
JO  - The Journal of Korean Institute of Communications and Information Sciences
PY  - 2024
DA  - 2024/1/1
DO  - 10.7840/kics.2024.49.5.671
KW  - LSTM
KW  - 802.11p
KW  - V2X communication
KW  - link adaptation
KW  - Buffer Overflow
KW  - PER
KW  - throughput
AB  - Recently, machine learning has improved prediction performance of rapidly changing channel environments and compensations in V2X communication. However, conventional prediction system based on received signals did not consider performance stabilization by deciding the sending method of a transmitter. Thus, we propose a mechanism that adopts optimal sending method derived from Long Short-Term Memory (LSTM) during link adaptation of the transmitter. For increasing throughput, we also propose minimum speed limiting and overflow prediction algorithm to address buffer overflow caused by high traffic speed. When applying the Modulation and Code Scheme (MCS) prediction algorithm, throughput increases by an average of 10% compared to the current method using delayed MCS prediction results. Moreover, MCS prediction performance improves when the vehicle becomes faster, and transmission interval decrease. Through using the proposed scheme, throughput increases by 2.5 times than the current method using minimum speed limitation algorithm, which has overflow problem, and Packet Error Rate (PER) decreases in the range of 5dB to 10dB SNR.