Conceitos essenciais
Batch array codes (BACs) can support the same type of requests as original batch codes but with reduced redundancy, by allowing storage nodes to perform local computations over the stored data.
Resumo
The paper studies an array code version of batch codes, called batch array codes (BACs). In a BAC, each storage node stores a bucket containing multiple code symbols and responds with a locally computed linear combination of the symbols in its bucket during the recovery of a requested symbol.
The key insights are:
BACs can support the same type of requests as original batch codes but with reduced redundancy. The authors establish information-theoretic lower bounds on the code lengths and provide several code constructions that confirm the tightness of the lower bounds for certain parameter regimes.
For general parameters, the authors prove a lower bound on the code length of BACs, showing that it is at least mn/(m-k+1), where n is the number of information symbols, k is the number of parallel requests, and m is the number of storage nodes. This bound is shown to be tight for certain cases.
The authors also provide improved lower bounds for the cases when k < m < 2k and m = k+2. These bounds are again shown to be tight by explicit code constructions.
The constructions demonstrate that BACs can achieve shorter code lengths compared to the original batch codes for the same scenarios, implying lower storage overhead.