Mobile Ad-Hoc Networks are increasingly important for tactical and disaster-relief operations, but remain highly vulnerable to Electronic Warfare attacks on the shared electromagnetic spectrum. The proprietary protocols that defend against this (Beechat’s HopSync being the contemporary example) require specialised FPGA radios and are not available for independent study, leaving open the question of whether the same architectural pattern is implementable on commodity hardware at all.

This thesis presents and empirically evaluates an Adaptive Frequency Hopping protocol that combines three mechanisms: an HMAC-SHA256 Time-based One-Time Password for stateless channel rendezvous between nodes, kernel-evidence-driven blacklist learning to suppress repeat visits to compromised channels, and a dual-radio “ping-pong” handover that masks the radio re-tune cost. The implementation runs on two Raspberry Pi 5s with off-the-shelf AR9271 USB Wi-Fi adapters; jamming is simulated via the Linux tc/netem traffic-control facility.

Across a 30-run campaign (six experimental configurations × five simulated time windows, ~720,000 packets per direction), the full design delivers 92.17 ±0.32 % packet delivery ratio under continuous-wave jamming against 54.90 ±0.13 % for the worst-case static baseline while maintaining consistent improvement over all baselines. The +37.27 percentage-point improvement decomposes additively into a +22.84 pp contribution from time-based hopping and a further +14.43 pp from the adaptive blacklist, with the dual-radio architecture contributing to a ~1 s reduction in 95th-percentile latency. Per-event recovery medians sit at one bucket (2 s) on both algorithmic and end-to-end axes, approximately two orders of magnitude faster than the static baseline. The reference implementation and analysis pipeline are released as the principal open-source artefact of the project.