LSTM-Based Multi-Link Prediction for mmWave and Sub-THz Wireless Systems
- Resource Type
- Conference
- Authors
- Shah, Syed Hashim Ali; Sharma, Manali; Rangan, Sundeep
- Source
- ICC 2020 - 2020 IEEE International Conference on Communications (ICC) Communications (ICC), ICC 2020 - 2020 IEEE International Conference on. :1-6 Jun, 2020
- Subject
- Communication, Networking and Broadcast Technologies
Logic gates
Handover
Robots
Drones
5G mobile communication
3GPP
Millimeter wave
LSTM
machine learning
cellular wireless
- Language
- ISSN
- 1938-1883
A key challenge in mmWave systems is the rapid variations in channel quality along different beam directions. MmWave links are highly susceptible to blockage and small changes in the orientation of the device or appearance of blockers can lead to dramatic changes in link quality along any given direction. Many low-latency applications need to accurately predict link quality from multiple directions and multiple cells. This paper presents a novel long short term memory (LSTM)-based method for predicting multi-directional link quality in mmWave systems. The method is validated on two problems: A realistic simulation of multi-cell link tracking in an environment with randomly moving human and vehicular blockers at 28 and 140 GHz, and beam prediction in a real indoor setting at 60 GHz.