A cyclic shift-based network encoding and decoding method, device and related components

Abstract

The invention discloses a network coding and decoding method, device and related components based on cyclic shift. The method includes: constructing a base matrix based on (n, k) parameters and preset coding criteria; The bit structure is used to obtain a shift matrix; the original data is obtained, and the original data is divided into k source packets; k source packets are used as system packets, and k source packets are shifted and encoded m times according to the shift matrix to obtain m verification packets; combine the system packets and the verification packets to obtain n encoding packets, where n=k+m; the decoding process includes: randomly selecting k encoding packets from the n encoding packets, and then adding k encoding packets Find the encoding package with exposed bits in the package, obtain the exposed bits, and then substitute the exposed bits into other encoding packages, and then eliminate the exposed bits according to the preset operation method; and so on, until all the k encoding packages The exposed bits are eliminated. The present invention can effectively reduce coding and decoding overhead through incremental difference and cyclic shift.

Description

technical field [0001] The present invention relates to the technical field of computer software, in particular to a cyclic shift-based network encoding and decoding method, device and related components. Background technique [0002] The core idea of ​​the distributed system is to obtain (n, k) combination properties (CombinationProperty, CP) through coding, that is, to encode k original independent tasks into n (n≥k), in the results of these n tasks, any k can restore the original k storage or calculation results. Shift-and-Add (SA) coding only includes shift and addition operations, which avoids a large number of multiplication operations in the linear coding stage and reduces computational complexity. Corresponding to SA is Zigzag Decoding (ZD). ZD can avoid a large number of complex operations such as multiplication and division, and only requires a simple back-substitution algorithm, which greatly reduces the computational burden. In addition, a code that has both th...

AI Technical Summary

Problems solved by technology

But this encoding still has some shortcomings. Inc-Diff encoding is asymmetric and when n and k are relatively close, its storage overhead is relatively large, and its storage overhead is m(k-1), where m=n-k

[0004] 2. In 2017, a coding scheme of ZD nature with a small overhead - cyclic shift permutation matrix coding was proposed, and this coding is called Cyc-Shift. This code has low coding and decoding complexity, and its storage overhead is relatively low. Inc-Dif should be small, for However, this code can only be applied to scenarios where n≤2k, and is not general

The Base-Shift code can be applied to any (n, k) scene, however, when the ratio of m / k is larger, the overhead of this code is also very large, which still brings a great burden to the communication cost

[0006] The disadvantage of the existing CP-ZD code is that the overall overhead is relatively large. When the coded packet m is larger, the overhead increases faster, and the communication cost is high, which is not suitable for the application of large-scale distributed systems.

Images