An evolutionary approach towards contact plan design for disruption-tolerant satellite networks
Graphical abstract
Introduction
Today, optical and radar images are acquired continuously from orbit as they enable better understanding and improved management of the Earth and its environment. In particular, orbiting networks of distributed satellite sensors are emerging as a mean to extend Earth observation missions revisit time and ground coverage [1]. In order to optimize space-to-ground data delivery in satellite networks, nodes can cooperatively pass sensed data among them before establishing the final communication with the receiving ground station. To this end, satellites shall rely on efficient network protocols and algorithms properly designed to operate over inter-satellite links (ISLs) [2].
In the Internet, protocol operations are largely based on instant flow of information between sending and receiver nodes. However, in a space flight mission environment, the highly varying communication ranges on which mobile nodes have to operate, compels to face several disruptive situations [3]. For example, Fig. 1 illustrates a network of satellites with polar orbits where connectivity among them can only be guaranteed over the pole as ISL distances are minimal. Previous work has proposed to constrain ISL distances by applying strict flight-formations [4]; but later analysis showed that disruptions could be handled by considering delay and disruption tolerant networking (DTN) architecture [5].
Originally proposed in [6], DTN protocols assumes no continuous connectivity throughout the network. In particular, lapses in wireless links may be routine, lengthy, and recurring and should not be interpreted as errors or unwanted changes in topology. Indeed, the interval of time during which data may be passed from one node to another over a link is defined as a contact. However, the impermanent nature of contacts compels the nodes to have local storage for temporary retention of data which cannot be forwarded immediately. As a result, information can flow between nodes through contacts in a Store-and-Forward fashion until reaching its final destination. This forwarding scheme is illustrated in Fig. 2, where node 1 sends data to node 3. Since there is no direct communication between the source and destination nodes, data needs to go through intermediate node 2. However, a sporadic link exists between nodes 2 and 3, which requires to store in-transit data on the local storage of node 2 until the contact with node 3 is available.
In general, episodes of communication between satellites and ground stations on Earth are typically scheduled weeks or months before they occur. Specifically, the beginning and end of each contact can be accurately computed from known orbital elements on ground [7]. As a result, all forthcoming network communication opportunities, including ISLs, can be imprinted in a contact plan (CP) which can be conveniently provisioned to DTN nodes in advance [8]. Then, as traffic flows in the network, on-board routing schemes can take advantage of this topological information to take efficient forwarding decisions.
Recent research has suggested to further adapt and optimize CPs in accordance with available on-board transceivers on each node [9]. This process became known as contact plan design (CPD) and has received increasing attention from the community as different criteria can significantly impact the final network performance [10], [11], [12], [13], [14], [15]. Among existing schemes, the Traffic-Aware Contact Plan (TACP) [15] was proposed in 2016 as a suitable scheme for satellite networks as it considers all possible parameters including the expected amount of traffic and its generation time. Indeed, data acquisitions from on-board instruments are also centrally scheduled by a mission control center on ground [16].
Nonetheless, TACP operating performance relies on a mixed integer linear programming (MILP) formulation whose computation complexity becomes intractable even for medium-sized networks. The latter becomes a critical issue in DTN satellite networks where CPs have to be timely provisioned to orbiting nodes. In order to guarantee bounded duration in the CPD cycle, this article contributes with an heuristic alternative to TACP based on evolutionary algorithms (TACP-EA). Even though a similar algorithm based on simulated annealing have recently been proposed for the CPD problem [14], it assumes as optimization criteria tailored for navigation constellations with unpredictable traffic. To the best of authors knowledge, this is the first time an heuristic approach is proposed for the CPD of Earth observation satellite networks with scheduled traffic.
This work is an archival quality version of the article [17] with a more detailed explanation on the algorithm and an extended and improved performance analysis. The article is structured as follows. In Section 2 we provide a general overview of the CPD problem and the design constraints to later describe TACP and discuss its computational limitations. In order to overcome the latter, we describe TACP-EA as an alternative approach in Section 3. Next, in Section 4 we evaluate the performance of TACP-EA in a Low Earth Orbit (LEO) satellite network example to finally draw the conclusions in Section 5.
Section snippets
Contact plan design overview
Colloquially, a contact can be defined as a communication opportunity between DTN nodes. In practice, the information encoded in a contact include source and destination node, start and end time, and expected data rate. Indeed, these parameters can be calculated on ground by means of precise orbital mechanics comprising position, range, and attitude (orientation of the spacecraft in the inertial system) [7]. Also, the resulting values such as transmission power, modulation, bit-error-rate, etc,
Genetic algorithm optimization
When using the MILP theoretical model of TACP to solve small DTN systems, an optimal solution might be obtained in reasonable time frame with traditional cutting-plane mechanisms available in free solvers such as GNU Linear Programming Kit (GLPK) [23]. Nevertheless, when considering a sufficiently large instance of the problem, these methods fail to deliver an output in reasonable time as the required computing power increases exponentially with the size of the decision variables (satellites,
Performance analysis
In this section we describe a particular and realistic case study of a sample satellite network in order to analyze and compare the performance of the proposed TACP-EA with the theoretical model of TACP. As previously stated, to the best of author knowledge, TACP-EA is the first heuristic approach towards traffic-aware CPD, there is no other heuristic method that could be considered in this comparison.
The case study is comprised of a 4 polar-orbit DTN Low Earth Orbit (LEO) satellite network
Conclusion
In this work we addressed the problem of the CPD for networked satellite systems based on DTN. This type of networks can be particularly optimized by exploiting their predictable nature in order to take efficient planning decisions in advance. Among this, the design of the forthcoming communication opportunities imprinted in a CP was considered for nodes with constrained resources. To this end, we recalled TACP: a theoretical formulation based on a MILP statement which quickly becomes
References (27)
- et al.
Routing-aware fair contact plan design for predictable delay tolerant networks
Ad-Hoc Netw.
(2015) - et al.
Traffic-aware contact plan design for disruption-tolerant space sensor networks
Ad Hoc Netw.
(2016) - et al.
Analysis of the contact graph routing algorithm: Bounding interplanetary paths
Acta Astronaut.
(2012) - et al.
Wireless sensor systems for space and extreme environments: a review
Sens. J. IEEE
(2014) - et al.
Fractionated space infrastructure for long-term earth observation missions
- et al.
Space Mission Analysis and Design, vol. 8 of Space Technology Library
(1999) - et al.
Satellite formation, a mobile sensor network in space
- et al.
Delay and disruption tolerant networking (DTN): an alternative solution for future satellite networking applications
Proc. IEEE
(2011) A delay-tolerant network architecture for challenged internets
Fundamentals of Astrodynamics and Applications
(2007)
Application of contact graph routing to LEO satellite DTN communications
Design challenges in contact plans for disruption-tolerant satellite networks
IEEE Commun. Mag.
Cost-efficient topology design problem in time-evolving delay-tolerant networks
Cited by (14)
Routing in the Space Internet: A contact graph routing tutorial
2021, Journal of Network and Computer ApplicationsCitation Excerpt :The resulting contact plan comprises the envelope within which network connectivity can occur. That plan can then be post-processed to accommodate operational plans (anticipated episodes of disconnection due to power management, body-fixed instrument pointing, etc.) and to enhance fairness (Fraire et al., 2014a), adapt to mandated routing (Fraire and Finochietto, 2015a), accommodate known traffic flows (Fraire et al., 2016a, 2017c), mitigate congestion (Madoery et al., 2018a), reduce energy consumption (Fraire et al., 2018a, 2019b) or fit specific missions (Zhou et al., 2017). Contact plan design (Fraire and Finochietto, 2015b) is a distinct research area, out of scope of this tutorial.
On the Performance of Epidemic DTN Routing on Low-Power and Lossy Networks
2023, Proceedings - 2023 2nd International Conference on Electronics, Energy and Measurement, IC2EM 2023Toward Autonomous Cooperation in Heterogeneous Nanosatellite Constellations Using Dynamic Graph Neural Networks
2023, Proceedings of the International Astronautical Congress, IACDownlink aware data scheduling with delay guarantees in resource-limited leo satellite networks
2021, Peer-to-Peer Networking and Applications