Network environment-adaptive erasure coding scheme for decentralized ubiquitous storage
Gang Song, Quan Zhang, Xin Xu · 2023
The adaptive erasure coding scheme for decentralized ubiquitous storage is investigated in this paper to improve the level of durability and reduce repair bandwidth cost and repair time. First, the extended parameters Reed-Solomon (RS) codes are proposed, and the relationship equations relating to durability, repair bandwidth cost, encoding parameters and peers fluctuation rate are set. Subsequently, the Environment-adaptive Parameters Algorithm (EAPA) is proposed. In a variety of network environments, EAPA can adjust encoding parameters adaptively to retain the desired level of durability without increasing storage cost. The Optimal Repair Cost Parameters Algorithm (ORCPA) is proposed for different scenarios, which leads to the optimal repair bandwidth cost and achieves a desired level of durability. ORCPA adapts parameters in accordance with the ratio of slow peers in the network to reduce the repair time. Experiments show that EAPA and ORCPA are capable of significantly increasing level of durability and reducing repair bandwidth cost and repair time compared with several erasure codes with wide application in multi-peer failure scenarios.