| YAMEOGO KISWENDSIDA ARISTIDE CHILDERIC | M, 2回目発表 | サイバーレジリエンス構成学 | 門林 雄基 | 安本 慶一 | 妙中 雄三 |
|
title: Controller-Level Intrusion Detection for Bluetooth Low Energy Wearable Devices
abstract: Wearable devices increasingly rely on Bluetooth Low Energy (BLE), yet our security assessment of three commercial wearables shows that most vulnerabilities arise not from the BLE standard itself, but from weak vendor implementations. Observed issues include legacy pairing, plaintext data transmission, static MAC addresses, and insecure OTA mechanisms. To address this, we developed SecureWear-FW, a secure-by-default firmware framework enforcing LESC+MITM pairing, TLS 1.3 presets, signed OTA, secure storage, and automated misconfiguration checks. We further designed a lightweight on-device IDS that learns normal behavior directly on the microcontroller using multi-mode profiling and statistical anomaly detection with <20 KB RAM overhead. A systematic literature review confirms the feasibility of ML-based IDS on constrained devices but highlights the lack of BLE- and wearable-focused studies. Building on these findings, our next step is to create an OASIS-inspired BLE dataset and integrate a lightweight ML model into SecureWear-FW. language of the presentation: English | |||||