Novel Repair-by-Transfer Codes and Systematic Exact-MBR Codes with Lower Complexities and Smaller Field Sizes

Abstract

The $(n,k,d)$ regenerating code is a class of $(n,k)$ erasure codes with the capability to recover a lost code fragment from other $d$ existing code fragments. This paper concentrates on the design of exact regenerating codes at Minimum Bandwidth Regenerating (MBR) points. For $d=n-1$, a class of $(n,k,d=n-1)$ Exact-MBR codes, termed as repair-by-transfer codes, have been developed in prior work to avoid arithmetic operations in node repairing process. The first result of this paper presents a new class of repair-by-transfer codes via congruent transformations. As compared with the prior works, the advantages of the proposed codes include: i) The minimum of the finite field size is significantly reduced from $\left(\genfrac{}{}{0ex}{}{n}{2}\right)$ to $n$. ii) The encoding complexity is decreased from $n^4$ to $n^3$. As shown in simulations, the proposed repair-by-transfer codes have lower computational overhead when $n$ is greater than a specific constant. The second result of this paper presents a new form of coding matrix for product-matrix Exact-MBR codes. The proposed coding matrix includes a number of advantages: i). The minimum of the finite field size is reduced from $n-k+d$ to $n$. ii). The fast Reed-Solomon erasure coding algorithms can be applied on the Exact-MBR codes to reduce the time complexities.

References in corpus (5)

• Distributed Storage Codes with Repair-by-Transfer and Non-achievability of Interior Points on the Storage-Bandwidth Tradeoff
• Distributed Data Storage with Minimum Storage Regenerating Codes - Exact and Functional Repair are Asymptotically Equally Efficient
• On the Existence of Optimal Exact-Repair MDS Codes for Distributed Storage
• Fractional Repetition Codes for Repair in Distributed Storage Systems
• Exact Minimum-Repair-Bandwidth Cooperative Regenerating Codes for Distributed Storage Systems