MDLBA-LORA : A MULTIDIMENSIONAL LEARNING-BASED LIGHTWEIGHT ROUTING PROTOCOL FOR SECURE AND EFFICIENT COMMUNICATION IN LORAWAN.

Abstract

LoRa WAN deployments still suffer from unreliable routing, energy constraints, and other vulnerabilities to DoS, injection, jamming, and Man-in-the-Middle attacks. Against this backdrop, this work proposes an intelligent and secure routing framework called MDLBA-LoRa (Multidimensional Learning Based Algorithm for LoRa), which is developed upon lightweight fuzzy logic, a BRNN-inspired anomaly detector, cluster-based priority scheduling, and symmetric encryption with AES-128. MDLBA-LoRa incorporates dynamic reconfiguration through continuous fuzzy trust updates, real-time attacker list management, priority-based queue reordering, and adaptive routing that avoids malicious or low-trust nodes based on evolving network conditions. In this paper, the system is implemented in NS-2, utilizing an LoRa -specific agent that emulates sensor nodes, gateways, and network-server operations. Neighbor trust is evaluated based on real-time metrics such as RSSI, queue length, bandwidth usage, and processing delay using rule-based fuzzy scoring. In this regard, a bidirectional temporal evaluator classifies the traffic as either legitimate or malicious. On the other hand, online clustering identifies high-priority packets for immediate encrypted forwarding, buffering normal traffic for scheduled transmission. Duty-cycle emulation and priority-aware queueing are implemented to ensure realistic energy and delay behavior. Through experimental evaluation against multiple attack scenarios, it significantly outperforms the baseline DLBA, RLLoRa, and DLBA-LoRa frameworks. Packet delivery ratio reaches approximately 87.84%, throughput reaches up to 84 kbps, delay goes down to about 1.28 seconds, and energy consumption reduces to 0.083 J. The accuracy of intrusion detection is over 99.06%. Strong precision, recall, and F1 values are demonstrated. MDLBA-LoRa effectively ensures secure, reliable, and energy-efficient communication for large-scale IoT networks.