On Safeguarding Visible Light Communication Systems Against Attacks by Active Adversaries

We consider an indoor visible light communication (VLC) system comprising of a transmitter (Alice) and a receiver (Bob). The communication from Alice to Bob is prone to active attacks by a malicious node (Eve) installed nearby. Specifically, when the VLC channel contains idle slots, Eve launches impersonation attack-thus, the VLC channel becomes orthogonal multiple access (OMA). On the other hand, when the VLC channel remains fully occupied by Alice, Eve is left with no choice other than to transmit simultaneously and this constitutes a jamming attack-thus, the VLC channel becomes non-orthogonal multiple access (NOMA). To thwart the impersonation attack, Bob authenticates the received packets via binary hypothesis testing (BHT) by utilizing the channel gain as the transmit device fingerprint. As for the jamming attack, Bob treats Eve's interference as Gaussian noise to recover the data sent by Alice. We further study two different physical scenarios: a room and a corridor, respectively. In the former, Bob remains static, while in the latter, Bob is mobile. Thus, for the corridor scenario, Bob tracks Alice's channel via a linear Kalman filter whose prediction is then fed to the BHT as the ground truth. Simulation results show that: i) for OMA VLC, authentication becomes more effective as Eve's distance to Bob becomes more dissimilar to Alice's distance to Bob; ii) for NOMA VLC, the average decoding error probability increases approximately logarithmically with the increase in the interference power of Eve.