SERS: Secure & Efficient Random and Symbol Linear Network Coding Schemes

Recently, several security schemes for Random Linear Network Coding (RLNC) have been proposed to increase the immunity of the RLNC technology against security attacks. One of the presented security schemes aims at securing the Global Encoding Vectors (GEV) that use other vectors to maintain the proper encoding process of RLNC at intermediate nodes. However, this approach introduces overhead in terms of computational complexity (block cipher with multiple rounds and operations) and communication (2xn elements for each packet instead of n). To that end, this paper proposes a new scheme, $S E R S$, that overcomes the disadvantages and limitations of the existing security schemes by relying on a single GEV instead of two, which is the case of the original RLNC. The proposed scheme reduces the required computational complexity by eliminating AES encryption and by keeping the source RLNC encoding step as a secret. SERS is based on a dynamic key structure, and the introduced modifications result in a modern lightweight, and secure RLNC while achieving higher efficiency and minimizing the space of vulnerabilities. SERS exhibits minimal computational complexity and communication overhead, and it ensures message confidentiality and availability, in addition to source authentication when a homomorphic keyed hash function is employed. A second variant of the proposed scheme is also presented. The main advantages of the proposed scheme are that 1) it operates at the sub-generation level, 2) it can be implemented in parallel, and 3) it increases the security level by using different RLNC encoding matrices instead of just one.