Energy-Efficient Fast Data Retrieval Strategy Based on RS Coded Placement in LEO Constellation
Zhineng Wu,Shushi Gu,Zhikai Zhang,Qinyu Zhang,Yifeng Jin,Lei Zhang,Wei Xiang
DOI: https://doi.org/10.1109/vtc2024-spring62846.2024.10683149
2024-01-01
Abstract:Low earth orbit (LEO) constellation network is a crucial component and paradigm for big data applications in the future satellite Internet. However, the characteristic of multi-hop transmission significantly increases the data retrieval delay and energy consumption depending on the inter-satellite communications. To mitigate this issue, this paper explores encoding redundancy by reed-solomon (RS) codes to generate the data and parity blocks, and store them in the satellite nodes of LEO constellation. We propose a fast data retrieval strategy, involving three stages: data collection, parity block placement, and data retrieval. Through the derivations of delay and energy consumption, we find that both are intensively related to the placement of parity blocks. To reduce energy consumption during the data retrieval process, we formulate a minimum problem under the delay constraint, which is an integer nonlinear programming problem. Then, we design an energy-efficient parity block placement based on genetic algorithm (PBP-GA), which is heuristic with fast convergence property. Simulation results show that, PBP-GA achieves a comprehensive performance improvement in average data retrieval delay and total energy consumption, compared to other placement schemes, i.e., random placement, retrieval cluster priority placement, nearby placement and non-coding. Specifically, PBP-GA can find the optimal number of parity blocks in a practical scenario of data retrieval in LEO constellation.