Improving System Reliability by Joint Usage of Hash Function Bits and Error Correction Coding

Abstract

Hash functions are widely deployed in many cryptographic applications, such as message integrity checks, digital signatures, message authentication codes (MACs), and many other applications. In fact, hash functions are mainly dedicated to detect tampering and prove of message authenticity. Thus, if the hash digest does not match the sent digest, this indicates modification in the data or in the hash digest itself. This may be a result of transmission error or a possible abusing attack on the system. It is observed that most practical systems deploy a bit-size for the hash mappings which ranges from 160 bits to 512 bits or more. This is often over dimensioned to comply with standardized hash functions. Therefore, the hash function bits as output of pseudorandom mapping can be replaced partially and temporarily by some forward error correcting code ECC leaving the remaining bits for authentication without significant loss of authentication security. This work evaluates a practical example combining a simple interleaved Reed-Solomon single-byte error correcting code to replace a small part of the hash bits in the sent message. On the receiving side, the ECC bytes are used to correct errors and then removed and replaced by the remainder of the original hash value. The non-replaced large hash bits would contribute to let the miss-correction performance of the code approaches zero. This technique would reduce non-significantly or even negligibly the authentication level of the system, however would improve the overall system reliability in a great deal. No changes in the size and format of the existing authenticated message are necessary. The reliability improvement and security degradation in our proposed technique are evaluated and compared with the original designed values. Experimental simulations are also compared with those statistically computed under idealized randomizing assumptions for the hash function.