• Dr.M.K.Demde
• Shruti Kharbade
• Vaishnavi Pillay
• Leena Barange
• Rakhi Sarode

Reliable data exchange is essential in IoT deployments, yet a faulty device or a degraded wireless link can halt it. This paper reports a prototype that heals its own communication paths with the help of machine learning, built on ESP32 boards, Wi-Fi, MQTT and Python. The boards send periodic status reports covering signal strength, uptime, free memory and connection state to a Python backend through an MQTT broker, where a trained model rates each node's health. Dijkstra's algorithm proposes candidate paths, and each is scored on predicted health, signal quality, delay, hop count and cost. When a node disappears or a link weakens, the backend switches to the best remaining path and informs the affected nodes. Text and images are transferred with acknowledgments and resending of lost chunks, and a web dashboard presents node state, routes, predicted health and recovery events live. Dashboard observations from the five-node prototype confirm that monitoring, health scoring and route selection work together.

ESP32, IoT, machine learning, self-healing network, MQTT, dynamic routing, network monitoring.

"AI-Based Self-Healing Communication Network", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 10, page no.a744-a750, October-2026, Available :http://www.jetir.org/papers/JETIR2610081.pdf

"AI-Based Self-Healing Communication Network", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 10, page no. ppa744-a750, October-2026, Available at : http://www.jetir.org/papers/JETIR2610081.pdf