WiFi RSSI Passenger Movement Dataset: Boarding, Alighting, and Static Classes in a Simulated Bus Environment

This dataset contains temporal WiFi Received Signal Strength Indicator (RSSI) sequences collected for passenger movement classification in a public transport scenario. Experiments were conducted in a controlled indoor environment simulating a bus and bus stop, using a single WiFi access point positioned at the vehicle doorway. Data was collected from six mobile devices across four manufacturers to introduce hardware variability.

Each sample consists of 10 consecutive RSSI measurements recorded at 1 Hz over a 10-second observation window, corresponding to one of four movement classes: AA (remaining inside the vehicle), BB (remaining at the bus stop), AB (alighting — transitioning from inside to outside), and BA (boarding — transitioning from outside to inside).

The dataset contains 1,356 samples (approximately 340 per class) labeled with the movement class and a noise indicator distinguishing isolated-condition samples (single active device, n=160) from noisy-condition samples (four simultaneous devices, n=1,196). The balanced class distribution enables direct application of standard classification metrics. RSSI values range from −80 dBm to −19 dBm.

This dataset was used to benchmark 38 machine learning classifiers. The best result was achieved by a Gaussian Process classifier (accuracy: 81.6%, MCC: 0.756 on the combined dataset), with KNN (k=5) reaching MCC 0.907 on isolated-only samples. The dataset is intended to support reproducible research in RSSI-based sensing, automated passenger counting, and intelligent transportation systems.