Edge Computing and Communication for Energy-Efficient Earth Surveillance with LEO Satellites

Abstract

Modern satellites deployed in low Earth orbit (LEO) accommodate processing payloads that can be exploited for edge computing. Furthermore, by implementing inter-satellite links, the LEO satellites in a constellation can route the data end-toend (E2E). These capabilities can be exploited to greatly improve the current store-and-forward approaches in Earth surveillance systems. However, they give rise to an NP-hard problem of joint communication and edge computing resource management (RM). In this paper, we propose an algorithm that allows the satellites to select between computing the tasks at the edge or at a cloud server and to allocate an adequate power for communication. The overall objective is to minimize the energy consumption at the satellites while fulfilling specific service E2E latency constraints for the computing tasks. Experimental results show that our algorithm achieves energy savings of up to 18% when compared to the selected benchmarks with either 1) fixed edge computing decisions or 2) maximum power allocation.

DCT-Based Air Interface Design for Function Computation

“…), part of the computation will be displaced to the transmitter side in order to reduce the amount of transmitted data, latency, energy consumption, among many other metrics. Thus, the system has to find a trade off between the offloaded and the locally processed data [6][7][8][9]. This work is part of the project IRENE (PID2020-115323RB-C31), funded by MCIN/AEI/10.13039/501100011033.…”

Satellite Edge Computing for Real-Time and Very-High Resolution Earth Observation

“…Nevertheless, the research on SMEC to reduce the amount of data transmission and energy consumption in Earth observation applications is still in its infancy. Specifically, most of the literature on SMEC, with a few clear exceptions [3], [4] that include our previous work [6], deals with tasks generated at the mobile users and, hence, mimics the traditional functionality of a terrestrial mobile edge computing (MEC) but substituting the terrestrial edge nodes with satellites [7]. Currently, SMEC approaches for Earth observation can be divided into: early discard [3], [4] and compression [6].…”

“…Specifically, most of the literature on SMEC, with a few clear exceptions [3], [4] that include our previous work [6], deals with tasks generated at the mobile users and, hence, mimics the traditional functionality of a terrestrial mobile edge computing (MEC) but substituting the terrestrial edge nodes with satellites [7]. Currently, SMEC approaches for Earth observation can be divided into: early discard [3], [4] and compression [6]. Early discard of images with relatively low information content (e.g., cloud coverage) effectively reduces the amount of data transmitted to ground [3], [4].…”

“…In our previous work [6], we focused on an Earth observation scenario with power allocation for communication and where the source satellite chooses between directly downloading the data or compressing it locally. The present paper represents a major extension of our previous work and a major deviation with respect to the SMEC literature described above by considering real-time high-resolution Earth observation applications where the real-time requirements are defined by the dynamics of the system.…”

“…Given that the satellites have a limited battery supply, the objective of distributing the tasks across the satellites in the constellation is to minimize the overall energy consumption while fulfilling the limitations of the processing frequency at the satellites' CPU, and the rates at the ISLs and satellite-to-ground link. The present work is a major extension of our previous study [6], where we only considered local edge computing and cloud computing, a single and homogeneous task at the satellites, a fixed CPU frequency, and a fixed compression ratio. In the present paper, we consider the optimization with distributed processing, where the set of satellites participating in the processing, the CPU frequency, and the compression ratio are optimization variables.…”