Performance Evaluation of Secure Network Coding Using Homomorphic Signature

TL;DR: Assessment of the feasibility of implementing Homomorphic Network Coding in an off-the-shelf laptop/smartphone platform and evaluation of the processing and delay performance when such implementations are deployed in a simple network scenario.

Abstract: Network coding has gained significant attention by improving throughput and reliability in disruptive MANETs. Yet, it is vulnerable to attacks from malicious nodes. In order to prevent malicious attacks, we have explored the use of secure network coding schemes based on homomorphic properties of cryptographic systems. While homomorphic methods, especially those based on public key cryptography, provide strong protection against both external and internal attacks, they do increase processing overhead as they require complex cryptographic operations (e.g., exponentiation, multiplication, modular operations). The goal of this paper is two fold: assess the feasibility of implementing Homomorphic Network Coding in an off-the-shelf laptop/smartphone platform, and evaluate the processing and delay performance when such implementations are deployed in a simple network scenario. To this end, we have implemented in LINUX an RSA-based homomorphic algorithm which exhibits very competitive processing efficiency as compared with published (public-key) schemes. For the LINUX implementation we have measured the processing delay for various flow and parameter settings. We have then integrated the homomorphic processing model (with associated O/H) in a MANET network simulator. Using this simulator, we have evaluated the performance of secure network coding under various network conditions and have compared it with other secure network coding schemes. We conclude the paper with a discussion of secure coding feasibility and cost for different application scenarios.