Hybrid Security Framework and Machine Learning Based Anomaly Detection for Machine-to-Machine Communications
B Satyanarayana Murthy, Raja Rao PBV, M Prasad, Kiran Sree Pokkuluri, P. J. R. Shalem Raju, K.S. Kumar · 2025
Recently, both IOT as well as Industry 4.0 have augmented the evolution of Machine-to-Machine(M2M) Communication. This has instigated a rush in the expanse of data directed by devices in submissions like smart cities, self-driving automobiles, as well as factory automation. However, safeguarding the sanctuary of M2M data transmission postures hard problems since M2M devices are controlled in terms processing power, memory, and energy. While outdated methods of encryption can be operative, they are often moreover expensive in terms of overhead for such resource constrained devices. To resolve these limitations, the projected research develops an cohesive security building using lightweight encryption systems together with machine learning based intrusion detection systems (IDS) for effectual data security.The proposed method adopts lightweight encryption procedures to guarantee energy effective encryption schemes like PRESENT, and SPECK. In addition, machine learning models like decision tree and neural networks are made to perceive anomalies in M2M communications, which supports the scheme to detect threats in real time much better. The anticipated resolution comprises two levels of security which are built on an intelligent threat discovery scheme and network-based encryption throughout data transfer. This paper outfits self-defined simulations and metrics to demonstrate that the hybrid method is operative in dropping energy feeding, minimizing latency, while still being able to recognize erudite attacks like man-in-the-middle attack and replay attacks inside resource limited contexts.The incorporation of machine learning with lightweight encryption delivers an operative, flexible, and scalable resolution to defence M2M communications across numerous segments while guaranteeing confidentiality and integrity of sensitive information and still summit the rigorous requirements linked with constrained devices. This paper demonstrates the effectiveness of the proposed hybrid solution compared to the existing traditional approach to security and presents a framework for upcoming studies which focus on enhancing machine learning model application for real-time M2M security purposes.