Feistel Network Assisted Dynamic Keying based SPN Lightweight Encryption for IoT Security

Abstract

In the last few years Internet-of-Things (IoT) technology has emerged significantly to serve varied purposes including healthcare, surveillance and control, business communication, civic administration and even varied financial activities. Despite of such broadened applications, being distributed, wireless based systems, IoTs are often considered vulnerable towards intrusion or malicious attacks, where exploiting the benefits of loosely connected peers, the attackers intend to gain device access or data access un authentically. However, being resource constrained in nature while demanding time-efficient computation, the majority of the classical cryptosystems are either computationally exhaustive or limited to avoid attacks like Brute-Force, Smart Card Loss Attack, Impersonation, Linear and Differential attacks, etc. The assumptions hypothesizing that increasing key-size with higher encryption round can achieve augmented security often fail in IoT due to increased complexity, overhead and eventual resource exhaustion. Considering it as limitation, in this paper we proposed a state-of-art Generalized Feistel Network assisted Shannon-Conditioned and Dynamic Keying based SSPN (GFS-SSPN) Lightweight Encryption System for IoT Security. Unlike classical cryptosystems or even substitution and permutation (SPN) based methods, we designed Shannon-criteria bounded SPN model with Generalized Feistel Network (SPN-GFS) modelthat employs 64-bit dynamic key with five rounds of encryption to enable highly attack-resilient IoT security. The proposed model was designed in such manner that it could be suitable towards both data-level security as well as device level accesscredential security to enable a "Fit-To-All" security solution for IoTs. Simulation results revealed that the proposed GFS-SSPN model exhibits very small encryption time with optimal NPCR and UACI. Additionally, correlation output too was found encouragingly fair, indicating higher attack-resilience.

Construction of a Security Defense Model for the University's Cyberspace Based on Machine Learning

“…Tonkal et al proposed a machine learning method with neighborhood component analysis for detecting DDoS attacks in software defined networks [8]. Gurumanapalli et al [9] proposed a state-of-the-art generalized Feistel networkassisted Shannon condition and dynamic key based SSPN (GFS-SSPN) lightweight encryption system for IoT security to achieve high attack resilience. Majeed et al [10] used intelligent machine learning methods to design IoT assisted drones.…”