Estimating Congestion in Train Cars by Using BLE Signals

Abstract

In many of the world’s major cities, commuter trains provide vital transportation support and thus play an essential role in our daily lives. Therefore, it has become necessary to estimate the degree of congestion in each train car, both to improve passenger comfort levels and, more recently, to prevent worsening the COVID-19 pandemic infection rate. However, it is difficult to estimate the degree of congestion within a train without violating passenger privacy. The same issues are true for busses, which is noteworthy because we have previously developed and evaluated a system that can estimate the degree of congestion within a bus while protecting passenger privacy by using Bluetooth Low Energy (BLE) signals. In this paper, we report on our efforts to extend that system to railway use, which were conducted on actual trains in cooperation with Kintetsu Railway Co., Ltd. During this trial, we collected BLE signals and used the data to estimate congestion levels in each car using an ML regression model. The results show that the mean absolute error (MAE) and the mean absolute percentage error (MAPE) could be estimated at accuracy levels of 5.56 and 0.27, respectively.