A method for constructing multi-satellite collaborative strategy for space-based Internet of Things based on element spatiotemporal extension graph

Abstract

The present invention discloses a method for constructing a multi-satellite collaborative strategy for a space-based Internet of Things based on an element spatiotemporal extension graph, comprising the following steps: constructing a weighted spatiotemporal connection graph (STCG) to perform steady-state processing on a satellite dynamic network; constructing an element weighted spatiotemporal connection graph (ESTCG) based on the weighted spatiotemporal connection graph (STCG) to perform integrated management of device resources, link resources, storage resources, and computing power resources in a satellite network; and constructing a multi-satellite collaborative strategy for a space-based Internet of Things based on the element weighted spatiotemporal connection graph (ESTCG). The present invention adds a resource management mode that can achieve integrated management of device resources, link resources, storage resources, and computing power resources in a satellite network; with the help of the element spatiotemporal extension graph, the motion laws and inherent resources of a satellite cluster are transformed into intuitively displayed static graphs, simplifying the formulation process of the multi-satellite collaborative strategy for a space-based Internet of Things, making the multi-satellite collaborative strategy for a space-based Internet of Things simpler, more direct, and more efficient.

Description

Technical Field

[0001] The present invention relates to the field of multi-satellite collaboration technology in a space-based Internet of Things, and in particular to a method for constructing a multi-satellite collaboration strategy for a space-based Internet of Things based on an element spatiotemporal extension graph. Background Art

[0002] In my country, the computing power industry has become a core area of ​​innovation and development in the information sector. In recent years, governments at all levels have intensively introduced a series of policies to support the computing power industry, viewing it as a key sector for future development. With the ongoing implementation of national computing industry strategies such as "Cloud Guizhou" and "Eastern Data West Computing," computing power has become a vital infrastructure and an engine for economic innovation and development. The industry's scale is expected to exceed 200 billion yuan by 2025.

[0003] The development of the computing power industry has driven the growth of the information industry, and with it, user demand for computing services continues to expand. The demand for computing power has gradually expanded from traditional urban environments to rural areas, and then to areas beyond the reach of traditional communication modes. This means that even in areas difficult to reach by mobile networks, such as oceans, deserts, and outer space, users still expect real-time, seamless, and efficient computing services to solve user needs such as trajectory prediction, target identification, and real-time data processing. To meet users' growing demand for computing power, satellite communications, as a communication technology that provides efficient coverage, has received significant attention in recent years and is considered a core component of 6G communication technology. In the future, high- and low-orbit satellite communications and terrestrial mobile communications are expected to seamlessly connect, forming an integrated three-dimensional network spanning air, space, land, and sea. As computing moves with the network, satellite-to-ground computing will integrate satellite systems, air networks, terrestrial communications, and cloud computing, creating an emerging computing architecture that will expand the scope of digital services. In its 2022 Top Ten Technology Trends Forecast, released in December 2021 by Alibaba's DAMO Academy, space-ground computing—integrated satellite and ground-based communications and computing—is a key future development trend, powerfully promoting the comprehensive digitalization of air, space, land, and sea. The number of low-orbit satellites is expected to explode over the next three years, with satellites and their ground systems becoming new computing nodes.

[0004] Currently, transmitting frontline data back to a ground-based cloud center via satellite is a relatively mature solution for carrying computing power. However, in this process, satellites are only used for transparent data forwarding. Users' computing services must go through multiple steps: uploading to the satellite - forwarding to the ground-based cloud center - transmitting the results back to the satellite - and then the satellite sending the results back to the user. This significantly increases the service latency of users' services and makes it difficult to meet their real-time business needs. To address this problem, the concept of "space computers" has been proposed, attempting to use satellites to directly provide computing services to users, which can greatly reduce business processing latency. However, because satellites generally have only limited computing and storage capabilities, they are easily affected by the space environment and may cause errors, and the computing and storage capabilities of a single satellite are relatively weak. Therefore, the concept of a space-based computing power-first network (CFN) has been further proposed, attempting to combine multiple satellites in a satellite constellation to achieve collaborative computing.

[0005] Space-based CFN refers to a new computing model that uses space-based satellite cluster network collaborative computing to provide all-weather, all-terrain, strong real-time, and seamless computing services to users on the ground, flying objects, space stations, and ships. Figure 1 As shown. Using space-based CFN, users can obtain real-time computing support at any time and any location around the world, realizing a true "space computer" and "space brain," enhancing the user experience. Once aerospace satellite clusters have the ability to collaboratively process complex tasks, the business and service models of ground users will change, expanding user application directions:

[0006] 1) Local processing of user tasks. For example, meteorological satellites can not only remotely sense and photograph targets of interest, but also perform computational steps such as cleaning, denoising, edge recognition, and target identification in the process of transmitting remote sensing images back. They can directly transmit information such as the type, location, and trajectory of the target of interest to the target unit, thus achieving rapid response to user tasks. At the same time, each step consumes only a small amount of satellite computing resources and energy resources, saving overall satellite resources.

[0007] 2) Fully utilize the transmission path to process information and achieve "processing while transmitting." For example, when using satellites for long-distance data transmission, an encryption-decryption step algorithm diagram can be constructed to implement the encryption and decryption process of the satellite during the transmission process, thus building a more secure data transmission path. Because each mission can be configured with a completely different "encryption-decryption" task, security is improved.

[0008] 3) Saving intersatellite link bandwidth. Supporting data cleaning during the step-by-step processing, data reduction is achieved after each step. Non-core data is cleaned and only the processed core data is forwarded, thus significantly saving satellite link resources.

[0009] Therefore, it is necessary to study collaborative computing in satellite clusters. However, due to the particularity of space-based satellite cluster networks, space-based CFN networks need to solve the distributed computing challenges brought about by the high dynamics and weak connections in space-based satellite constellations. In addition, with the development of microservice technology and network function virtualization, the traditional client-server model has been deconstructed, and server-side applications have been decoupled into atomic functional components. Different categories of functional components have different computing power and can handle different tasks. However, due to the limited computing and storage resources in a single satellite, it is difficult to deploy all types of functional components on a single satellite. Therefore, only some types of functional components are deployed in each edge server, which adds new difficulties to the management of computing resources in the satellite network.

[0010] The current status of research on satellite high dynamics and satellite computing power coordination strategies is as follows:

[0011] 1. High dynamic research of satellites

[0012] Thanks to recent breakthroughs in aerospace technology, the highly dynamic nature of satellite topology, its vast coverage, and short transmission distances to the Earth have made the deployment of collaborative computing on space-based systems a new trend in the development of communication systems. Research on high-dynamic satellite topology is divided into two main parts: 1) The establishment of highly dynamic satellite network topology, focusing on the establishment of satellite network topology based on the periodic changes in satellite orbital position under high-speed satellite motion; and 2) The time-varying nature of satellite links, focusing on the time-varying link capacity of satellite-to-ground links and inter-satellite links. Due to the periodicity and predictability of satellite topology, when studying the dynamic topology of satellite networks, topology control strategies are generally used to mask the dynamic nature of the topology. Currently, satellite network topology control strategies mainly include virtual topology strategies, virtual node strategies, and coverage domain partitioning methods. A typical example of a virtual topology strategy is the "snapshot" concept. The concept of "snapshot" was first proposed by Gounder et al. (Gounder VV, Prakash R, Abu-Amara H. Routing in LEO-based Satellite Networks[C]. 1999 IEEE Emerging Technologies Symposium. Wireless Communications and Systems (IEEE Cat. No. 99EX297). IEEE, 1999: 22.1-22.6.). Fischer et al. later formalized the concept of "snapshot" (Fischer D, Basin D, Engel T. Topology Dynamics and Routing for Predictable Mobile Networks[C]. 2008 IEEE International Conference on Network Protocols. IEEE, 2008: 207-217.). To describe the changing patterns of satellite network snapshots, Wang et al. proposed theoretical formulas for calculating the number and length of snapshots in Low Earth Orbit (LEO) satellite networks.The concept of virtual nodes was first proposed by Mauger et al. (Mauger R, Rosenberg C. QoS Guarantees for Multimedia Services on a TDMA-based Satellite Network [J]. IEEE Communications Magazine, 1997, 35(7): 56-65.), and was subsequently popularized and applied by Ekici et al. (Ekici E, Akyildiz IF, Bender MD. A Distributed Routing Algorithm for Datagram Traffic in LEO Satellite Networks [J]. IEEE / ACM Transactions on networking, 2001, 9(2): 137-147.). In 2013, Lu et al. described the main features of the virtual node strategy and conducted a formal description and optimization design (Lu Yong, Zhao Youjian, Sun Fuchun. Satellite Network Routing Technology [J]. Journal of Software, 2014(05).). The concept of coverage domain partitioning was proposed by Hashimoto et al. (Hashimoto Y, Sarikaya B. Design of IP-based Routing in a LEO Satellite Network [C]. Proceedings of Third International Workshop on Satellite-Based Information Services. New York: ACM, 1998: 81-88.). It divides the Earth into multiple evenly distributed cellular regions, each of which is supported by the nearest satellite. Virtual topology strategies and coverage domain partitioning have fewer constellation restrictions, but they result in more topological variations and a significant additional computational load. Virtual node strategies are applicable to a limited constellation range, but they offer the advantages of a fixed topology and no additional computational load. Space-based computing networks are primarily based on space-based satellite constellations; therefore, further modeling of the time-varying nature of satellite constellation links is necessary. For satellite-to-ground links, due to the weak arrival signals from satellites and the complex ground conditions, the received signal is often a superposition of multipath signals from all directions.The literature (Lu Y, Sun F, Zhao Y. Virtual Topology for LEO Satellite Networks Based on Earth-Fixed Footprint Mode [J]. IEEE communications letters, 2013, 17(2): 357-360.) demonstrates through mathematical analysis that the envelope of a multipath signal with a direct component follows the Rice distribution; the Rice channel model is currently the most commonly used channel model for satellite-to-Earth links. For intersatellite links, since intersatellite link loss is primarily free-space propagation loss, its magnitude is primarily related to the intersatellite distance. References (Liu Zhongrong. Research and Implementation of Wireless Mobile Channel Models [D]. University of Electronic Science and Technology of China, 2004) and Zhou Yunhui. Research on Satellite Network QoS Routing Protocol and Its Optimization Theory [D]. Tsinghua University, 2007) provide detailed modeling of dynamic satellite networks, but use fixed values ​​to describe the transmission rate of inter-satellite links, failing to reflect the time-varying nature of inter-satellite link capacity. Reference (Wang Yahui. Capacity Analysis and Access Control Strategy Research for Non-geostationary Satellite Communication Systems [D]. University of Electronic Science and Technology of China, 2018) links transmission loss with inter-satellite distance, deriving a relationship between received signal power and inter-satellite distance. Therefore, in the study of highly dynamic satellite networks, inter-satellite link capacity can be calculated using the time-varying inter-satellite distance and the conclusions of reference (Xue R, Yu H, Cheng Q. Adaptive Coded Modulation Based on Continuous Phase Modulation for Inter-Satellite Links of Global Navigation Satellite Systems [J]. IEEE Access, 2018, 6:20652-20662).

[0013] 2. Research on satellite computing power coordination strategies

[0014] With the continuous development of space-based satellite networks based on satellite communication technology, inspired by edge computing (MEC), introducing MEC into satellite networks is a good way to provide computing services to ground users. Reference (J.Liu, X.Du, J.Cui, M.Pan and D.Wei, "Task-Oriented Intelligent Networking Architecture for the Space–Air–Ground–Aqua Integrated Network," in IEEE Internet of Things Journal, vol. 7, no. 6, pp. 5345-5358, June 2020, doi: 10.1109 / JIOT.2020.2977402.) extends edge computing to the Space-Air-Ground-Aqua Integrated Network (SAGAIN), treating satellites as edge computing nodes and providing the necessary computing and caching capabilities. References (Q. Tang, Z. Fei, B. Li and Z. Han, "Computation Offloading in LEO Satellite Networks With Hybrid Cloud and Edge Computing," in IEEE Internet of Things Journal, vol. 8, no. 11, pp. 9164-9176, 1 June 2021, doi: 10.1109 / JIOT.2021.3056569.) A dual-layer satellite MEC (SMEC) structure was developed in the LEO satellite network, and an optimization problem was formulated with the goal of minimizing the total energy consumption of ground users. Reference (H.Tan,M.He,T.Xia,X.Zheng and J.Lai,"A Novel Multi-level Computation Offloading Scheme at LEO Constellation Broadband Network Edge,"2020IEEE World Congresson Services(SERVICES),2020,pp.281-286,doi:10.1109 / SERVICES48979.2020.00062.) proposes a multi-level computation offloading scheme that can provide computation offloading options for ground users.Shangguan Boyi et al. (Shangguan Boyi, Liu Wei, Le Peng, Wang Mi, Jiang Hao, Yan Zheren. A satellite-ground collaborative computing migration method for space information networks [J]. Journal of Wuhan University (Information Science Edition), 2019, 44(03): 459-466. DOI: 10.13203 / j.whugis20170269.) proposed a satellite-ground collaborative computing migration method for space information networks. This method divides the scheduling optimization model into two parts: a data transmission model and a computing migration scheme selection model, solving the problem that the system's remote sensing data transmission capacity changes with the network topology. Zhang Keke et al. (Zhang Keke, Sun Yukun, Xia Lei, Zhu Zhencai, Wang Jing. A method for scheduling distributed collaborative tasks on-orbit for network satellites [J]. Journal of Harbin Engineering University, 2019, 40(02): 393-399.) designed a "Sandwich" space information network architecture. The network transmission efficiency is suitable for implementing on-orbit distributed collaborative computing under multi-task requirements. Based on this, a distributed disconnection and reconnection algorithm is proposed to overcome the computing resource limitations of a single satellite and improve on-orbit computing capabilities. Zhixuan Tang et al. (Z. Tang, H. Zhou, T. Ma, K. Yu and X. S. Hen, "Leveraging LEO Assisted Cloud-Edge Collaboration for Energy Efficient Computation Offloading," 2021 IEEE Global Communications Conference (GLOBECOM), 2021, pp. 1-6, doi:10.1109 / GLOBECOM46510.2021.9685309.) proposed a low-orbit satellite-assisted terrestrial satellite network (TSN) architecture for cloud-edge collaborative computing offloading. The problem of minimizing the energy consumption of the entire TSN under the constraint of Quality of Service (QoS) was constructed. The user association scheme and task scheduling strategy were optimized through Deep Neural Networks (DNN). Then, the Successive Convex Approximation (SCA) algorithm was used to solve the optimized offloading ratio, computing resource allocation and transmission power to achieve optimal energy consumption.Haofei Li et al. (H.Li, C.Chen, C.Li, L.Liu and G.Gui, "Aerial Computing Offloading by Distributed Deep Learning in Collaborative Satellite-terrestrial Networks," 2021 13th International Conference on Wireless Communications and Signal Processing (WCSP), 2021, pp. 1-6, doi: 10.1109 / WCSP52459.2021.9613173.) proposed a satellite-terrestrial collaborative network distributed offloading algorithm (Collaborative satellite-terrestrial Network Distributed Offloading, CNDO) based on parallel neural networks, which can effectively reduce the search space of offloading strategies, avoid the dimensionality curse, and give multiple optimal offloading decisions to ensure users' computing needs and low energy consumption and low latency requirements. Zhu Xiangming et al. (Zhu Xiangming, Liu Shanyun, Yang Bin, Qi Yinan, Zhang Xingming. A distributed collaborative computing migration method for satellite-ground collaborative communication system [P]. Zhejiang Province: CN113709775B, 2022-01-07.) fully utilize the two-layer edge computing architecture in the satellite-ground collaborative communication system, and avoid the two-way propagation delay caused by the centralized decision-making process through a distributed migration mechanism, thereby significantly reducing system delay, solving the problem of long decision-making delay in the centralized migration strategy in the satellite-ground collaborative communication system, and ensuring user delay-sensitive services.

[0015] Nichoas et al. proposed a neural network computing framework based on the Internet of Things (IoT). This framework maps the neural network's neuron operations to IoT nodes and leverages IoT satellite communication to transmit data between neurons, reducing transmission latency and improving information privacy. Machine learning has also been applied to the development of computing power coordination strategies. Kim et al. (SIKim and HSKim, "A research on dynamic service function chaining based on reinforcement learning using resource usage," in the 2017 Ninth International Conference on Ubiquitous and Future Networks (ICUFN). IEEE, 2017, pp. 582–586.) calculated the physical and virtual resources used by each network function and then used a reinforcement learning algorithm based on a reward-penalty paradigm to construct a dynamic service function chain. Wang et al. (T. Wang, J. Zu, G. Hu, and D. Peng, “Adaptive Service Function Chain Scheduling in Mobile Edge Computing via Deep Reinforcement Learning,” IEEE Access, vol. 8, pp. 164-922–164-935, 2020) proposed a deep reinforcement learning approach to solve the computing power coordination strategy problem. This approach can detect changes in the edge computing system environment and make adaptive adjustments to service requests. Fu et al. (X. Fu, F.R. Yu, J. Wang, Q. Qi, and J. Liao, “Service function chain embedding for NFV-enabled IoT based on deep reinforcement learning,” IEEE Communications Magazine, vol. 57, no. 11, pp. 102–108, 2019) proposed an algorithm based on experience replay and target network based on deep reinforcement learning.Qin et al. (Y. Qin, Q. Xia, Z. Xu, P. Zhou, A. Galis, OF Rana, J. Ren, and G. Wu, “Enabling Multicast Slices in Edge Networks,” IEEE Internet of Things Journal, vol. 7, no. 9, pp. 8485–8501, 2020) proposed an efficient heuristic algorithm based on reinforcement learning (RL) to study the problem of latency-oriented network slicing with a given latency guarantee level. In addition, Topcuoglu et al. (H. Topcuoglu, S. Hariri and Min-You Wu, “Performance-effective and low-complexity task scheduling for heterogeneous computing,” in IEEE Transactions on Parallel and Distributed Systems, vol. 13, no. 3, pp. 260–274, March 2002) proposed the HEFT algorithm. Due to its low complexity and high performance, HEFT has also been widely used in computing power coordination strategies.

[0016] In summary, existing research on satellite cluster collaborative computing technologies faces challenges with the high dynamics, weak connectivity, and computational resource management of satellite networks. Furthermore, the process for formulating multi-satellite collaborative strategies for space-based IoT is complex and inefficient. Combining existing research on distributed technologies based on computational routing and integrating them with spatiotemporal extended graph technology, there is an urgent need to research distributed collaborative computing methods for inter-satellite clusters in space-based networks. Summary of the Invention

[0017] The purpose of the present invention is to provide a method for constructing a multi-satellite collaborative strategy for a space-based Internet of Things based on an element spatiotemporal extension graph with strong computing power and high performance. On the premise of solving the high dynamics, weak connection and computing power resource management problems of satellite networks, the method provides services to users by jointly participating in the calculation of multiple computing nodes and concentrating computing power.

[0018] The technical solution to achieve the purpose of the present invention is: a method for constructing a multi-satellite collaborative strategy for a space-based Internet of Things based on an element spatiotemporal expansion graph, comprising the following steps:

[0019] Step 1: Construct a weighted spatiotemporal connection graph (STCG) to stabilize the satellite dynamic network.

[0020] Step 2: Based on the weighted spatiotemporal connectivity graph (STCG), an element weighted spatiotemporal connectivity graph (ESTCG) is constructed to perform integrated management of device resources, link resources, storage resources, and computing resources in the satellite network.

[0021] Step 3: Based on the element-weighted spatiotemporal connectivity graph (ESTCG), a multi-satellite collaboration strategy for the space-based Internet of Things is constructed.

[0022] Furthermore, in step 1, the weighted space-time connection graph STCG is constructed. The conditions for constructing the satellite inter-satellite link include the satellite position, geometric visibility conditions, antenna visibility conditions, and channel capacity requirements, where:

[0023] (1) Satellite location

[0024] Based on the WGS-84 coordinate system: the origin coincides with the center of the Earth, the Z axis points to the CTP direction of the Earth's poles as defined by BIH 1984.0, the X axis points to the intersection of the zero meridian plane of BIH 1984.0 and the CTP equator, and the Y axis, Z axis, and X axis form a right-handed coordinate system;

[0025] The position of the satellite is determined by the six major Kepler parameters. The semi-major axis a and the eccentricity e determine the shape of the satellite orbit. The satellite orbit plane is determined by the inclination δ and the right ascension Ω of the ascending node. The perigee angular distance ω determines the position of the orbit on the orbital plane, and the true anomaly f determines the position of the satellite in the orbit.

[0026] Among the six Kepler parameters that describe the satellite's undisturbed motion, only the true anomaly is a function of time, while the others are constants. Therefore, the calculation of the satellite's instantaneous position lies in calculating the true anomaly.

[0027] Assuming that satellite v i The six Kepler parameters are The subscript i represents the satellite number. To calculate the true anomaly, two auxiliary parameters are introduced, namely the eccentric anomaly E and i Peace anomaly M i , where M i The expression of time variation is:

[0028]

[0029] Where G is the gravitational constant, M is the mass of the Earth, Satellite v i The time of passing the perigee, t is the time of observing the satellite; according to Kepler's equation, the eccentric anomaly angle E i for

[0030] E i (t) = M i (t)+e i sin E i (t) (2)

[0031] Eccentric anomaly E i Solve it by iterative method and get the true perigee f i :

[0032]

[0033] Satellite v i The distance to the center of the earth is

[0034] r i (t) = a i (1-cosE i (t)) (4)

[0035] Then satellite v i The coordinates in the WGS-84 coordinate system (x i (t),y i (t),z i (t)) is

[0036]

[0037] Therefore, the satellite v i With v j The distance between

[0038]

[0039] (2) Geometric visibility conditions

[0040] Assuming the thickness of the atmosphere is h, when the communication link between two satellites is tangent to the atmosphere, this is the critical state of geometric visibility between satellites;

[0041] Two satellites v i and v j The following conditions must be met to establish an intersatellite link:

[0042] d ij (t)≥(R e +h) (7)

[0043] Among them, R e is the radius of the earth; the above formula is converted into a constraint on the satellite distance:

[0044]

[0045] (3) Antenna visibility conditions

[0046] Assuming that the satellite antennas have sufficient tracking capabilities, when two satellites are in each other's antenna beam, they are considered to be visible between the satellite antennas. The satellite antennas are fixed to the ground, and the satellite antenna scanning range is considered to be a cone with the satellite antenna transmitting point as the vertex and the maximum scanning range of the antenna beam as the cone angle. The height of the cone is aligned with the line connecting the satellite antenna transmitting point and the center of the earth.

[0047] Assume satellite v i The beam scanning range is γ i , satellite v j The beam scanning range is γ j , under the condition that the satellite meets the geometric visibility, the four antenna visibility situations include: satellite v i and v j Both satellites are outside the scanning range of each other's beam, and both satellites are invisible; satellite v i In satellite v j Within the beam scanning range, but the satellite v j Not on Satellite V i Within the beam scanning range, the two satellites are still not visible; Satellite v i and satellite v j Both satellites are just within the scanning range of each other's beams, and the two satellites are just visible; satellite v i and satellite v j Normally visible;

[0048] When the antenna visibility condition is met, the center of the earth and the satellite v j 、Satellite v i The central angle θ of the circle should satisfy:

[0049]

[0050] Similarly, the above expression can be converted into the distance expression:

[0051]

[0052] (4) Channel capacity requirements

[0053] In natural space, satellite v i ,v j , the propagation loss between i≠j is

[0054]

[0055] in, is the operating wavelength, f is the carrier frequency, and c is the speed of light;

[0056] Based on the free space propagation formula, v i ,v j , the power of the received signal between i≠j is as follows:

[0057]

[0058] Among them, P e is the transmitting power of the transmitting source, G e is the transmission gain of the directional antenna, G r is the receiving antenna gain, and P e G e It is called the effective isotropic radiated power EIRP, and the signal-to-noise ratio at the receiving end is expressed as

[0059]

[0060] Where k = 1.379*10 -23 =-228.6dBW / Hz is the Boltzmann constant, T represents the noise temperature, N0 is the noise power spectrum density, and B is the bandwidth of the intersatellite link. According to the Shannon formula, the intersatellite channel capacity is

[0061] C ij (t) = Blog2(1+S ij (t)) (14)

[0062] Assume that the minimum communication channel capacity between satellites is C0, that is, C ij (t)>C0, the two satellites v i ,v j , i≠j can communicate at time t, otherwise not; then under the condition of signal power attenuation, satellite v i ,v j , i≠j can establish a satellite link only if

[0063]

[0064] Similarly, the above expression can be converted into the distance expression:

[0065]

[0066] v i ,v j , the expression of the intersatellite link status between i≠j changing with time is:

[0067]

[0068] Among them, g ij (t) = 1 represents satellite v i ,v j ,The link with i≠j is connected at time t, otherwise it is disconnected.

[0069] Furthermore, the stabilization process of the satellite dynamic network is performed as described in step 1, specifically as follows:

[0070] Based on the discrete topological sequence, time slot partitioning is applied to truncate satellite links with continuously changing states into periodically unchanged satellite link segments. The satellite's operating cycle T is divided into N consecutive and equal time slots, with each time slot lasting Δt = T / N. The satellite network is considered static within each time slot, and the topological structure is viewed as a static steady-state graph that connects multiple satellite nodes according to the existence of satellite inter-satellite links. In order to establish connections between multiple steady-state graphs and describe the global topological information of the satellite time-varying network, virtual links are established between the same satellite nodes in adjacent time slots, thus forming a virtual topology. The steady-state graphs are connected in chronological order to form a complete weighted spatiotemporal connection graph (STCG) of the satellite network throughout the entire cycle.

[0071] In order to find the routing strategy with the lowest delay in data transmission between satellites, the data transmission delay in the link and the storage delay in the node are used as the weights of the actual link and virtual link, respectively. That is, the data transmission delay in space is used as the weight of the actual link within the time slot, and the data transmission delay in time is used as the weight of the virtual link between time slots. Therefore, the weight of the weighted space-time connection graph STCG is also determined, and the weighted adjacency matrix of the weighted space-time connection graph STCG is written accordingly.

[0072] Furthermore, in step 1, the satellite dynamic network is stabilized, including the construction of the satellite topology within the time slot, the satellite topology between time slots, and the STCG, as follows:

[0073] (1) Satellite topology within a time slot

[0074] Assume that the available set of times at the beginning of each time slot is T = {t1, t2, ..., t N} means, where t k ∈T represents the time when the kth time slot starts, and t k+1 -t k =Δt; In order to represent the change of node status over time, suppose there are M nodes in the satellite network and the available set of nodes in the kth time slot is Indicates; v 编号 The form represents the satellite node itself, that is, The form only represents the nodes within a certain time slot;

[0075] In a single time slot, any two satellites v i ,v j i≠j in the kth time slot, that is, node The average link transmission rate between

[0076]

[0077] When transmitting data unit data (1 bit) in a satellite network, the data is transmitted between any two nodes in the kth time slot. The transmission delay between

[0078]

[0079] The weighted adjacency matrix of the satellite network topology at the kth time slot is expressed as

[0080]

[0081] represents the satellite v i ,v j ,i≠j is not connected in the kth time slot and data transmission is not possible;

[0082] (2) Satellite topology between time slots

[0083] Introduce virtual links between different replica nodes of the same satellite in adjacent time slots along the time growth direction, that is, nodes With node It is one-way connected in time; since the virtual link represents the topological connection of the nodes in time, there is no virtual link between the replica nodes of different satellites in different time slots, that is, when i≠j, the node With node Not connected;

[0084] A virtual link only represents the temporal continuity of the topology. A virtual link means that data is stored in a node and waits for transmission in the next transmission slot. Therefore, the weight of the virtual link should be the remaining time of the current slot when the data is transmitted to the node.

[0085] Assume that data arrives at the node The delay in the kth time slot is The data is at the node The transmission delay between

[0086]

[0087] The weighted adjacency matrix of the satellite network topology structure in adjacent time slots {k, k+1} is expressed as

[0088]

[0089] (3) Construction of STCG

[0090] The weighted adjacency matrix within and between time slots includes the entire process of topological connection, data transmission and storage between satellites. Therefore, all weighted adjacency matrices within the operating cycle [0, T) are combined in the order of time slot changes to record all information transmitted in the satellite dynamic network. The combined matrix is ​​the weighted adjacency matrix of STCG, as shown in the following formula:

[0091]

[0092] Among them, G N,1 It represents the periodicity of the dynamic topology changes of the satellite. Accordingly, the dynamic topology of the satellite network and the data transmission process in the satellite network are restored according to the Graph.

[0093] Furthermore, in step 2, the element weighted space-time connection graph ESTCG is constructed based on the weighted space-time connection graph STCG to perform integrated management of device resources, link resources, storage resources, and computing power resources in the satellite network, as follows:

[0094] (1) Nodes: The nodes of the element graph are divided into two categories: satellites and functional modules;

[0095] Functional modules are abstracted as nodes directly connected to satellites. Since the functional modules equipped by satellite nodes do not change over time, there is no need to consider time slot changes when discussing the feature graph;

[0096] Assume that the satellite network provides Num types of functional modules, and all the x-th, 1≤x≤Num types of functional modules are represented as an array in 1≤j≤M represents satellite v j Therefore, all functional module nodes in the satellite network can use F = {f 1 ,f 2 ,…,f Num Due to the limitations of computing resources and storage capacity, each edge satellite is only equipped with some functional modules, so some elements in the set F are empty.

[0097] (2) Edges: The edges in the feature graph are divided into two categories: edges between satellites or edges between satellites and functional modules;

[0098] The edge between the satellite and the functional module indicates that the satellite is equipped with a functional module. The edge from the satellite to the functional module indicates the process of the corresponding functional module processing data, and vice versa indicates the process of returning results.

[0099] Located in v j The data delay of the functional processing unit data of type x, 1≤x≤Num, that is, 1 bit, is expressed as

[0100]

[0101] where c jx (cycles / s) is v j The CPU frequency of the x type function; if v j There is no function of type x, c jx =0(cycles / s) and ct jx =∞(s / bit); therefore, the weight of the edge from the satellite node to the functional module of type x,1≤x≤Num is recorded as

[0102]

[0103] The function module is directly installed on the satellite, and the process of returning the result to the satellite no longer consumes time, and the weight of the edge from the function of type x,1≤x≤Num to the satellite node is

[0104]

[0105] Except for the edges between satellites, the edges from satellite to functional modules, and the edges from functional modules to satellites, there are no other edges in the element graph within a single time slot, and the weight of disconnected edges is infinite;

[0106] In k time slots, the weighted adjacency matrix of the satellite network represented by ESTCG is obtained:

[0107]

[0108] Where Inf represents a matrix whose diagonal elements are 0 and the rest of the elements are ∞;

[0109] In adjacent time slots {k, k+1}, the weighted adjacency matrix of the satellite network topology with added elements is shown as

[0110]

[0111] The weighted adjacency matrix of the satellite network after combining time slot division with feature graph is represented as

[0112]

[0113] Among them, GM N,1 It represents the periodicity of the dynamic topology changes of the satellite; accordingly, the dynamic topology of the satellite network and the transmission and processing process of data in the satellite network are restored according to GraphM.

[0114] Furthermore, in step 3, a multi-satellite coordination strategy for the space-based Internet of Things is constructed based on the element-weighted spatiotemporal connectivity graph ESTCG, as follows:

[0115] (1) Mission coordination between the two satellites

[0116] Assume that at time q, A pending service is initiated and a service with a workload of D bits is sent to the satellite v j In the process, the satellite v j The delay in processing this service is expressed as

[0117]

[0118] The input values ​​of the function shortest(G, t, v, f) are the weighted adjacency matrix G of the graph, the service start time t, the service starting point v, and the function module f as the end point;

[0119] Although ESTCG statically represents satellite networks as a weighted adjacency matrix GraphM, compared to quasi-static networks, it still has the following limitations when processing services:

[0120] First, due to the dynamic nature of satellite networks, the topological connectivity between two time slots may vary. Therefore, the total data transmission delay within a single time slot cannot exceed the time slot Δt, otherwise there is a risk of transmission interruption. Assuming that the weight of each connected link is 0.4Δt, within a time slot, data can only reach node v3 from node v1 and cannot reach node v5.

[0121] Secondly, the weight of the virtual link in the ESTCG model represents the process of data storage and waiting for transmission; the weight of the virtual link is not fixed. The weight of the virtual link between two time slots of a node is equal to the remaining time of the current time slot when the data arrives at the node;

[0122] In seeking When the optimal routing is between If the data from the node Transmit to node The meaning of this is that the data is temporarily stored in node v1 for a time slot, so the weight between time slots is is Δt; and the data passes through the time slot The weight of the time slot is 0.4Δt, then the data has spent 0.4Δt in the current time slot when it reaches node v2; and when it passes through the cross-time slot edge When , it means that the data is temporarily stored in node v2 for the remaining time of the kth time slot, that is, the time slot weight is 0.6Δt; The sum of the weights of the shortest routes between them is 1.4Δt;

[0123] When measuring service processing delay, it is necessary to add restrictions based on formula (30):

[0124] ① Assume any two satellites and When the amount of data transmitted is Dbit, The shortest delay of the optimal route transmission unit delay between For data to be successfully transmitted within a single time slot, the following conditions must be met:

[0125]

[0126] ② Since the weight of the cross-time slot edge is not fixed and is related to the time when the data arrives at the node, the weight of the cross-time slot edge needs to be calculated in real time based on the choice of data transmission path;

[0127] (2) Space-based Internet of Things mission coordination

[0128] Assume that at time q, A pending service is initiated. The data volume of this task is Dbit and needs to be processed by a function module of type x. Assume that the data volume ratio of the task to each satellite in the satellite network is {λ1,λ2,…,λ M}, that is, assigned to the functional module f x The amount of data is {λ1×D,λ2×D,…,λ M ×D}, the delay of space-based Internet of Things collaborative processing of this service is the delay of the latest task assigned to it in the functional module, and can be expressed as

[0129]

[0130] In order to reduce the processing delay of services as much as possible, the goal of multi-satellite collaboration in space-based IoT is to minimize the service processing delay. The optimization problem is then expressed as

[0131]

[0132] And the constraints must be met:

[0133] ① Assume any two satellites and When the amount of data transmitted is Dbit, The shortest delay for transmitting unit data through the best route between For data to be successfully transmitted within a single time slot, the following conditions must be met:

[0134]

[0135] ② Since the weight of the cross-time slot edge is not fixed, it is related to the time when the data arrives at the node. Therefore, the weight of the cross-time slot edge needs to be calculated in real time based on the choice of data transmission path;

[0136] The BPSO algorithm is used to solve this optimization problem. Assuming that the size of the particle swarm is U and I represents the maximum number of iterations of the swarm, the position and velocity of the u≤Uth particle in the i≤Ith iteration are as follows:

[0137]

[0138]

[0139] Where, The value in represents the proportion of data allocated to each satellite, that is, The actual meaning is λ j The value of The value of represents the particle moving step to find the optimal solution, and the value is between [0,1]. When i<I, particle u first tracks the historical optimal position p. ubest and the historical optimal position g of the entire particle swarm best Update speed, as shown below:

[0140]

[0141] Where μ is the inertia weight, γ1 and γ2 are acceleration factors, and β1 and β2 are random numbers uniformly distributed in the interval [0,1]. The direction updates the position as follows:

[0142]

[0143] symbol Indicates that the value is rounded up, and Represents the permutation operation between the elements of two matrices. The specific process is to first determine The index of the element with the largest value in is taken as the position to be replaced, and then the value is calculated according to the constraints in the above formula. The index of the element that is to be replaced with the position to be replaced is used to complete the replacement operation; finally, the fitness value is calculated by the following formula:

[0144]

[0145] In summary, the optimal inter-satellite collaborative computing solution is solved based on the BPSO algorithm.

[0146] Compared with the existing technology, the present invention has the following significant advantages: (1) It proposes an element spatiotemporal expansion graph. Compared with the method of managing satellite networks, it not only solves the problems of high dynamics and weak connections between satellites, but also adds a resource management mode, which can realize the integrated management of equipment resources, link resources, storage resources and computing power resources in the satellite network; (2) Based on the element spatiotemporal expansion graph, the space-based Internet of Things task coordination strategy is studied. With the help of the element spatiotemporal expansion graph, the movement law and intrinsic resources of the satellite cluster are transformed into a static graph for intuitive display, which simplifies the formulation process of the space-based Internet of Things multi-satellite coordination strategy. The space-based Internet of Things multi-satellite coordination strategy is simpler, more direct and more efficient; (3) On the premise of solving the problems of high dynamics, weak connections and computing power resource management of satellite networks, it is possible to provide services to users by concentrating computing power through multiple computing nodes that jointly participate in the calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0147] Figure 1 This is a schematic diagram of the space-based CFN network architecture.

[0148] Figure 2 This is a schematic diagram of satellite orbit parameters.

[0149] Figure 3 It is a schematic diagram of satellite geometric visibility.

[0150] Figure 4 This is a diagram of a satellite antenna.

[0151] Figure 5 This is a diagram showing satellite antenna visibility.

[0152] Figure 6 It is a time-varying slice diagram.

[0153] Figure 7 It is a virtual topology connection diagram.

[0154] Figure 8 It is a schematic diagram of the static expression of the dynamic topology of the STCG satellite network.

[0155] Figure 9 It is a schematic diagram of the satellite element map for a single time slot.

[0156] Figure 10 It is a schematic diagram of a multi-slot satellite element map.

[0157] Figure 11 This is a diagram of business processing in ESTCG.

[0158] Figure 12 This is a diagram showing an example of transmission interruption.

[0159] Figure 13 This is a schematic diagram of an example of cross-time slot routing.

[0160] Figure 14 This is a performance comparison chart of business processing methods. DETAILED DESCRIPTION

[0161] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0162] The present invention provides a method for constructing a multi-satellite collaborative strategy for a space-based Internet of Things based on an element spatiotemporal expansion graph, comprising the following steps:

[0163] Step 1: Construct a weighted spatiotemporal connection graph;

[0164] Step 2: Construct a weighted spatiotemporal connectivity graph of elements;

[0165] Step 3: Satellite collaboration technology based on ESTCG.

[0166] As a specific example, the construction of the weighted spatiotemporal connection graph described in step 1 is as follows:

[0167] Satellites can be categorized as low Earth orbit (LEO), medium Earth orbit (MEO), and geostationary Earth orbit (GEO) based on their orbital altitudes. Their relatively high-speed motion results in highly dynamic satellite network topologies, with link connectivity changing over time and the network frequently being disconnected. The time-varying nature of network topology results in relatively short link lifetimes and constantly changing connectivity between satellites. Therefore, shielding against satellite dynamics is a primary task in satellite computing collaboration. This paper proposes a time-varying graph of a weighted space-time connection graph (STCG) to address the problem of stabilizing dynamic satellite networks.

[0168] (1.1) Conditions for establishing intersatellite links

[0169] The existence of inter-satellite links is a necessary condition for information transmission and exchange between satellites. The existence of inter-satellite links represents whether the satellite nodes are connected. Therefore, before formulating the optimal routing between satellite networks, it is necessary to analyze the construction conditions of the satellite inter-satellite links.

[0170] 1. Satellite location

[0171] To analyze the visibility between satellites, we first need to determine the position of the satellites. The research in this paper is based on the WGS-84 coordinate system: the origin coincides with the center of the Earth, the Z axis points to the Conventional Terrestrial Pole (CTP) direction defined by BIH1984.0, the X axis points to the intersection of the zero meridian plane of BIH1984.0 and the CTP equator, and the Y axis, Z axis, and X axis form a right-handed coordinate system. The position of the satellite can be determined by the six major Kepler parameters, such as Figure 2 As shown, the semi-major axis a and eccentricity e determine the shape of the satellite orbit, the satellite orbit plane can be determined by the inclination δ and the right ascension of the ascending node Ω, the perigee angular distance ω determines the position of the orbit on the orbital plane, and the true anomaly f determines the position of the satellite in the orbit;

[0172] Among the six Kepler parameters that describe the satellite's irradiated motion, only the true anomaly is a function of time, while the others are constants. Therefore, the key to calculating the satellite's instantaneous position is to calculate the true anomaly. Assume that the satellite v i The six Kepler parameters (subscript i represents the satellite number) are In order to calculate the true anomaly, two auxiliary parameters need to be introduced: the eccentric anomaly E i Peace anomaly M i , where M i The expression of time variation is

[0173]

[0174] Where G is the gravitational constant, M is the mass of the Earth, Satellite v i The time of passing the perigee, t is the time of observing the satellite. According to Kepler's equation, the eccentric anomaly angle E i for

[0175] E i (t) = M i (t)+e i sin E i (t) (2)

[0176] Eccentric anomaly E i Usually, an iterative method is used to solve the problem. The specific solution process will not be described in detail in this invention. The true perigee f i :

[0177]

[0178] Satellite v i The distance to the center of the earth is

[0179] r i (t) = a i (1-cosE i (t)) (4)

[0180] Then satellite v i The coordinates in the WGS-84 coordinate system (x i (t),y i (t),z i (t)) is

[0181]

[0182] Therefore, the satellite v i With v j The distance between

[0183]

[0184] 2. Geometric visibility conditions

[0185] For two Earth satellites, they are geometrically visible only when they are both above a horizontal plane tangent to the Earth's atmosphere. Let the thickness of the atmosphere be h. When the communication link between the two satellites is tangent to the atmosphere, this is the critical state of geometric visibility between the satellites. Figure 3 As shown, the two satellites v i and v j The following conditions must be met to establish an intersatellite link:

[0186] d ij (t)≥(R e +h) (7)

[0187] Among them, R e is the radius of the Earth. The above formula can be converted into a constraint on the satellite distance:

[0188]

[0189] 3. Antenna visibility conditions

[0190] Satellite geometric visibility is a prerequisite for establishing an intersatellite link, but the scanning angle of the satellite antenna and its tracking capability are also important factors to be considered in building an intersatellite link. An intersatellite link can be established between satellites only when the connection line between the satellite antennas is within the overlapping area of ​​the two satellites' coverage area, and the relative angular velocity of the satellites is within the range that the antennas can track. The present invention assumes that the satellite antennas have sufficient tracking capabilities. When two satellites are in each other's antenna beam, the satellite antennas are considered to be visible. The satellite antennas are installed in a fixed position to the ground, such as Figure 4 As shown, the antenna scanning range of the satellite can be regarded as a cone with the satellite antenna transmitting point as the vertex and the maximum range that the antenna beam can scan as the cone angle, and its height is in the same straight line as the line connecting the satellite antenna transmitting point and the center of the earth.

[0191] Assume satellite v i The beam scanning range is γ i, satellite v j The beam scanning range is γ j , under the condition that the satellite meets the geometric visibility, Figure 5 Four antenna visibility situations are shown. Figure 5 As shown in (a), satellite v i and v j Both satellites are outside the scanning range of each other's beams and are not visible. Figure 5 As shown in (b), satellite v i In satellite v j Within the beam scanning range, but the satellite v j Not on Satellite V i Within the beam scanning range, the two satellites are still not visible; Figure 5 As shown in (c), satellite v i and satellite v j Both are just within the scanning range of the other party's beam, and the two satellites are just visible; Figure 5 As shown in (d), satellite v i and satellite v j Normally visible. When the antenna visibility conditions are met, the center of the earth and the satellite v j 、Satellite v i The central angle θ of the circle should satisfy:

[0192]

[0193] Similarly, the above expression can be converted into the distance expression:

[0194]

[0195] 4. Channel capacity requirements

[0196] As the distance of space transmission increases, the power attenuation of the signal increases, the channel capacity decreases, and the bit error rate rises sharply. Therefore, the design of actual intersatellite links must consider the impact of space transmission distance on channel capacity. i ,v j , the propagation loss between i≠j is

[0197]

[0198] in, is the operating wavelength, f is the carrier frequency, and c is the speed of light. Based on the free space propagation formula, v i ,v j , the power of the received signal between i≠j is as follows:

[0199]

[0200] Among them, Pe is the transmitting power of the transmitting source, G e is the transmission gain of the directional antenna, G r is the receiving antenna gain, and P e G e It is called the effective isotropic radiated power (EIRP), and the signal-to-noise ratio at the receiving end can be expressed as

[0201]

[0202] Where k = 1.379*10 -23 =-228.6dBW / Hz is the Boltzmann constant, T represents the noise temperature, N0 is the noise power spectrum density, and B is the bandwidth of the intersatellite link. According to Shannon's formula, the intersatellite channel capacity is

[0203] C ij (t) = Blog2(1+S ij (t)) (14)

[0204] Assume that the minimum communication channel capacity between satellites is C0, that is, C ij (t)>C0, the two satellites v i ,v j , i≠j can communicate at time t, otherwise it cannot. Under the condition of signal power attenuation, satellite v i ,v j , i≠j can establish a satellite link only if

[0205]

[0206] Similarly, the above expression can be converted into the distance expression:

[0207]

[0208] Combining the formula, we can get v i ,v j , the expression of the intersatellite link status between i≠j changing with time is:

[0209]

[0210] Among them, g ij (t) = 1 represents satellite v i ,v j ,The link with i≠j is connected at time t, otherwise it is disconnected.

[0211] 2. Satellite dynamic network topology steady-state processing method based on STCG

[0212] Although the relatively high-speed motion between satellites can lead to instability in intersatellite links and frequent changes in network topology, the fixed motion orbit makes its trajectory periodic and predictable. To shield the dynamic changes in the satellite network topology, the present invention, based on the concept of discrete topological sequence, uses time slot partitioning to truncate satellite links with continuously changing states into satellite link segments with stage-invariant states. The satellite's operating cycle T can be divided into N consecutive and equal-sized time slots, with the duration of each time slot being Δt = T / N. Within each time slot, the satellite network can be considered static, and its topology can be viewed as a static steady-state graph that connects multiple satellite nodes according to the existence of satellite intersatellite links. In order to establish connections between multiple steady-state graphs and thereby describe the global topology information of the satellite time-varying network, the present invention establishes virtual links between the same satellite nodes in adjacent time slots, thereby forming a virtual topology. The steady-state graphs are connected in chronological order to form a complete topology graph of the satellite network throughout the entire cycle - STCG.

[0213] The present invention aims to find a routing strategy that minimizes the latency for inter-satellite data transmission. To meet this requirement, the present invention uses the data transmission delay within a link and the storage delay within a node as the weights for actual and virtual links, respectively. Specifically, the data transmission delay in space is used as the weight for actual links within a time slot, and the data transmission delay in time is used as the weight for virtual links between time slots. This allows the weights of the STCG to be determined, allowing the weighted adjacency matrix of the STCG to be constructed accordingly.

[0214] 1. Satellite topology within the time slot

[0215] Since the connectivity of intersatellite links can be predicted by calculation, the topology of the satellite network in each time slot can be established based on the calculation results. The topology of the satellite network in a certain period of time can be divided into the topology of multiple time slots, such as Figure 6 As shown in Figure 2, the links between nodes indicate that the conditions for establishing inter-satellite links exist in space.

[0216] Assume that the available set of times at the beginning of each time slot is T = {t1, t2, ..., t N} means, where t k ∈T represents the time when the kth time slot starts, and t k+1 -t k =Δt. In order to represent the change of node status over time, we assume that there are M nodes in the satellite network and the available set of nodes in the kth time slot is Indicates. (Note: v 编号 The form represents the satellite node itself, that is, The form only represents the nodes within a certain time slot.)

[0217] In a single time slot, any two satellites v i ,vj i≠j in the kth time slot, that is, node The average link transmission rate between

[0218]

[0219] Therefore, when transmitting data unit data (1 bit) in the satellite network, the data is transmitted between any two nodes in the kth time slot. The transmission delay between

[0220]

[0221] Therefore, the weighted adjacency matrix of the satellite network topology at the kth time slot can be expressed as

[0222]

[0223] like represents the satellite v i ,v j ,i≠j is not connected in the kth time slot and data transmission is not possible.

[0224] Inter-slot satellite topology

[0225] The topology of the satellite network changes over time, and the link connection conditions between different time slots are different. In order to avoid link interruption during data transmission, when the data cannot be transmitted within the remaining time of the link stability, the data should be temporarily stored in the node and wait for transmission in the next time slot. The present invention considers the process of data storage and waiting for transmission as virtual transmission of data in time. Therefore, the present invention introduces virtual links between different replica nodes of the same satellite in adjacent time slots along the time growth direction, that is, nodes With node It is one-way connected in time. Since the virtual link represents the topological connection of the nodes in time, there is no virtual link between the replica nodes of different satellites in different time slots, that is, when i≠j, the node With node Absolutely not. Figure 7 It should be noted that the virtual link does not actually exist; it only represents the continuity of the topology in time.

[0226] The meaning of a virtual link is that data is stored in a node and waits for transmission in the next transmission time slot. Therefore, the weight of the virtual link should be the remaining time of the time slot when the data is transmitted to the node. Therefore, the weight of the virtual link is related to the routing strategy. Only when the path before the data reaches a node is determined, the weight of the virtual link between the node and the same node in the next time slot can be determined. Assuming that the data arrives at the node The delay in the kth time slot is The data is at the node The transmission delay between

[0227]

[0228] Therefore, the weighted adjacency matrix of the satellite network topology structure in adjacent time slots {k, k+1} can be expressed as

[0229]

[0230] Construction of STCG

[0231] The schematic diagram of the steady-state processing process of satellite dynamic network topology based on STCG is as follows: Figure 8 As shown, at this time the entire satellite network has been transformed into a steady-state topology, shielding the dynamics of the satellite network.

[0232] The weighted adjacency matrix within and between time slots includes the entire process of topological connection, data transmission and storage between satellites. Therefore, all weighted adjacency matrices within the operating cycle [0, T) are combined in the order of time slot changes to record all information transmitted in the satellite dynamic network. The combined matrix is ​​the weighted adjacency matrix of STCG, as shown in the following formula:

[0233]

[0234] Among them, G N,1 It represents the periodicity of the satellite’s dynamic topology changes. Accordingly, the dynamic topology of the satellite network and the data transmission process in the satellite network can be restored based on the graph.

[0235] As a specific example, the construction of the weighted spatiotemporal connectivity graph of elements in step 2 is as follows:

[0236] The construction of STCG achieves the stabilization of satellite networks and solves the problem of inter-satellite communication. The heterogeneous functional modules equipped in satellites contain a variety of computing resources. There is an urgent need for an effective way to manage and schedule computing resources to lay the foundation for collaborative computing between satellites. In this section, we further propose the element-weighted space-time connection graph (ESTCG) based on the STCG to complete the management of computing resources on the basis of achieving the stabilization of satellite networks. ESTCG includes various satellite network resources, such as satellite nodes, connectivity between satellites, and functional modules equipped on satellites. This section will introduce ESTCG from the perspectives of nodes and edges.

[0237] Satellite element diagram for a single time slot is as follows: Figure 9 As shown:

[0238] (1) Nodes: The nodes of the element diagram are divided into two categories: satellites and functional modules.

[0239] For the sake of intuition, the functional modules are abstracted as nodes directly connected to the satellite. Since the functional modules equipped by the satellite nodes do not change over time, there is no need to consider time slot changes when discussing the feature graph. Assuming that the satellite network can provide Num types of functional modules, all the xth, 1≤x≤Num types of functional modules can be represented as an array in 1≤j≤M represents satellite v j Therefore, all function module nodes in the satellite network can be represented by F = {f 1 ,f 2 ,…,f Num Due to the limitations of computing resources and storage capacity, it is not feasible to deploy all functional modules in one edge satellite. That is, each edge satellite is only equipped with some functional modules, so some elements in the set F are empty.

[0240] (2) Edges: The edges in the feature graph can be divided into two categories: edges between satellites or edges between satellites and functional modules.

[0241] Edges between satellites represent intersatellite links between them, and these edges represent the data transmission process. Edges between satellites and functional modules represent the functional modules installed on satellites. Edges from satellites to functional modules represent the data processing process of the corresponding functional module, and vice versa, the return process of the results. Weights between satellites have been discussed in previous sections; in this section, we will focus on the weights between satellites and functional modules.

[0242] Located in v j The data delay of the functional processing unit data (1 bit) of type x, 1≤x≤Num can be expressed as

[0243]

[0244] where c jx (cycles / s) is v j The CPU frequency of the x type function. If v j There is no function of type x, c jx =0(cycles / s) and ct jx =∞(s / bit). Therefore, the weight of the edge from the satellite node to the functional module of type x,1≤x≤Num can be recorded as

[0245]

[0246] Since the functions are installed directly on the satellite, the process of returning the results to the satellite does not consume time. In addition, the weight of the edge from the function of type x,1≤x≤Num to the satellite node is

[0247]

[0248] Aside from the edges between satellites, satellites to functional modules, and functional modules to satellites, there are no other edges in the element graph within a single time slot, and the weight of disconnected edges is infinite. Therefore, within k time slots, we can obtain the weighted adjacency matrix of the satellite network represented by ESTCG.

[0249]

[0250] Inf represents a matrix whose diagonal elements are 0 and the rest of the elements are ∞.

[0251] To overcome the dynamic nature of satellite networks, satellite networks are divided into multiple static graphs by time slots. After applying the ESTCG model, the satellite network can be represented as follows: Figure 10 Satellite element diagram for multiple time slots shown.

[0252] Since the functional modules configured in the satellite do not change over time, there is no need to consider the change of time slots when the functional modules process services. Due to the stability of the functional modules, time slot switching does not affect the processing of services by the functional modules, so the delay cost of time slot switching between functional modules is 0. Therefore, in adjacent time slots {k, k+1}, the weighted adjacency matrix of the satellite network topology with added elements can be expressed as

[0253]

[0254] The weighted adjacency matrix of the satellite network after combining time slot division with factor graph can be represented as

[0255]

[0256] Among them, GM N,1 It represents the periodicity of the satellite’s dynamic topology changes. Accordingly, GraphM can be used to recover the satellite network’s dynamic topology and the data transmission and processing processes in the satellite network.

[0257] As a specific example, the ESTCG-based satellite collaboration technology described in step 3 is as follows:

[0258] 3.1 Research on mission coordination between the two satellites

[0259] The coordination between satellite clusters can be decomposed into multi-step coordination between two satellites. Therefore, the realization of coordination between any two satellites is the core content of satellite coordination technology. When satellites coordinate to process tasks, it is necessary to consider not only the transmission delay of data between satellites, but also the delay of satellite processing data. Information such as the transmission delay of unit data between satellites, the calculation delay between satellites and computing modules, and the data return delay from computing modules to satellites are all included in the matrix GraphM. At this time, the various delays in processing services are abstracted as data at the same level, so they can be directly added, such as Figure 11 shown.

[0260] Assume that at time q, A pending service is initiated and a service with a workload of D bits is sent to the satellite v j In the process, the satellite v j The processing (computation and transmission) delay of this service can be expressed as

[0261]

[0262] The input values ​​of the function shortest(G, t, v, f) are the weighted adjacency matrix G of the graph, the service start time t, the service starting point v, and the functional module f as the end point. Its essence is the shortest path between nodes v and f in the network topology graph G, and t is used to determine the further positions of nodes v and f in the matrix G.

[0263] Although ESTCG statically represents the satellite network as a weighted adjacency matrix GraphM, compared to quasi-static networks, there are still some restrictions when processing services, as shown below:

[0264] First, due to the dynamic characteristics of the satellite network, there are differences in the topological connection between the two time slots. Therefore, the total data transmission delay within a single time slot cannot exceed the time slot Δt, otherwise there will be a risk of transmission interruption. To illustrate the problem more vividly, the present invention Figure 12 Give an example. Figure 12 As shown in the figure, assuming that the weight of each connected link is 0.4Δt, then within a time slot, data can only reach node v3 from node v1, but cannot reach node v5.

[0265] Secondly, unlike traditional static graphs, where the weights of all edges represent data transmission latency, the weights of virtual links in the ESTCG model represent the process of data storage and waiting for transmission. Therefore, the weights of virtual links are not fixed. The weight of a virtual link between two time slots at a node is equal to the remaining time in the current time slot when the data arrives at that node.

[0266] like Figure 13 As shown, in the When the optimal routing is between If the data from the node Transmit to node It means that the data is temporarily stored in node v1 for a time slot, so the weight between time slots is is Δt. The data passes through the time slot The weight of the time slot is 0.4Δt, then the data has spent 0.4Δt in the current time slot when it reaches node v2; and when it passes through the cross-time slot edge When , it means that the data is temporarily stored in node v2 for the remaining time of the kth time slot, that is, the time slot weight is 0.6Δt. Obviously, The sum of the weights of the shortest routes between them is 1.4△t.

[0267] Therefore, when measuring service processing delay, it is necessary to add restrictions based on formula (30):

[0268] (1) Assume any two satellites and When the amount of data transmitted is Dbit, they The shortest delay of the optimal route transmission unit delay between For data to be successfully transmitted within a single time slot, the following conditions must be met:

[0269]

[0270] (2) Since the weight of the cross-time slot edge is not fixed, it is related to the time when the data arrives at the node. Therefore, the weight of the cross-time slot edge needs to be calculated in real time based on the selection of the data transmission path.

[0271] 3.2 Research on Multi-satellite Mission Collaboration of Space-based Internet of Things

[0272] In order to improve data processing efficiency, the services generated by a single satellite often require the collaborative processing of multiple satellites. Based on the research on the collaboration between two satellites, we further studied the collaborative processing of multi-satellite tasks of the space-based Internet of Things and proposed the optimization method of multi-satellite collaboration of the space-based Internet of Things. Assume that at time q, A pending service is initiated. The data volume of this task is Dbit and needs to be processed by a function module of type x. Assume that the data volume ratio of the task to each satellite in the satellite network is {λ1,λ2,…,λ M}, that is, assigned to the functional module f x The amount of data is {λ1×D,λ2×D,…,λ M ×D}, the delay of multi-satellite collaborative processing of the service by the space-based Internet of Things is the delay of the latest task assigned to it in the functional module, and can be expressed as

[0273]

[0274] In order to reduce the processing delay of services as much as possible, the goal of multi-satellite collaboration in space-based IoT is to minimize the service processing delay. The optimization problem can be expressed as

[0275]

[0276] And the constraints must be met:

[0277] (1) Assume any two satellites and When the amount of data transmitted is Dbit, they The shortest delay for transmitting unit data through the best route between For data to be successfully transmitted within a single time slot, the following conditions must be met:

[0278]

[0279] (2) Since the weight of the cross-time slot edge is not fixed, it is related to the time when the data arrives at the node. Therefore, the weight of the cross-time slot edge needs to be calculated in real time based on the selection of the data transmission path.

[0280] The essence of the optimization problem (33) is an NP-hard problem. The complexity of its solution increases rapidly with the increase in the number of low-orbit satellites and the amount of mission data. Therefore, deterministic algorithms are not suitable for solving such problems. The binary particle swarm optimization (BPSO) algorithm is a popular group intelligent search algorithm for binary optimization problems. Compared with deterministic algorithms, it is more efficient and has a shorter execution time. Therefore, the present invention uses the BPSO algorithm to solve this optimization problem. We assume that the size of the particle swarm is U, and I represents the maximum number of iterations of the swarm. Then the position and velocity of the u≤Uth particle in the i≤Ith iteration are respectively as follows:

[0281]

[0282]

[0283] In the above formula, The value in represents the proportion of data allocated to each satellite, that is, The actual meaning is λ j The value of . The value of represents the particle moving step to find the optimal solution, and its value is also between [0,1]. When i<I, particle u first tracks its historical optimal position p ubestand the historical optimal position g of the entire particle swarm best Update its speed as shown below,

[0284]

[0285] Where μ is the inertia weight, γ1 and γ2 are acceleration factors, and β1 and β2 are random numbers uniformly distributed in the interval [0,1]. The direction updates its position according to the following formula,

[0286]

[0287] symbol Indicates that the value is rounded up, and Represents the permutation operation between the elements of two matrices. The specific process is to first determine The index of the element with the largest value in is taken as the position to be replaced, and then the value is calculated according to the constraints in the above formula. The index of the element that is replaced with the position to be replaced is used to complete the replacement operation; finally, the fitness value is calculated by the following formula.

[0288]

[0289] In summary, the optimal inter-satellite collaborative computing solution can be solved based on the BPSO algorithm.

[0290] Example 1

[0291] This embodiment will verify that the satellite collaboration solution based on ESTCG is superior to traditional single-satellite processing (taking satellite v1 as an example), cloud computing and other data processing methods. The simulation platform used in the present invention is MATLAB, in which 7 satellite nodes in the Starlink constellation are adopted as the test satellite networking structure of the present invention. The computing power of the computing nodes (including satellites and clouds) used in the present invention is shown in Table 1, where the business is initiated by satellite v1, the cloud is equipped with all types of functional modules, and each satellite is randomly equipped with some types of functional modules:

[0292] Table 1

[0293] equipment <![CDATA[Satellite v1]]> <![CDATA[Satellite v2]]> <![CDATA[Satellite v3]]> <![CDATA[Satellite v4]]> <![CDATA[Satellite v5]]> <![CDATA[Satellite v6]]> <![CDATA[Satellite v7]]> cloud Processing capacity cycles / s 150 200 180 170 130 150 170 500

[0294] pass Figure 14It can be seen that the multi-satellite collaborative processing capability of the space-based Internet of Things (SBIOT) remains optimal. When the data volume is small, the processing capability of a single satellite outperforms cloud computing. However, as the data volume increases significantly, cloud computing begins to outperform a single satellite. This is because multi-satellite collaboration combines the computing power of all satellites to process the service, and the transmission latency between satellites is relatively low, resulting in consistently optimal performance. When the data volume reaches approximately 610MB, the processing latency of multi-satellite collaborative processing increases slightly due to cross-slot collaboration during service processing. Single-satellite processing, on the other hand, does not require additional transmission latency, but its computing capability is relatively low, making it suitable only for services with relatively low data volumes. In particular, if the functional module for processing this service type is not configured on this single satellite, the service cannot be processed. For cloud computing, the relatively long distance between cloud computing and satellites results in relatively high transmission latency for data transmission from the satellite to the cloud. When processing tasks with relatively small data volumes, the transmission latency can even exceed the computational latency, resulting in significant resource waste. However, the computing power of cloud computing platforms is very strong. As the amount of data increases, the advantages of cloud computing gradually become apparent, so its performance begins to outperform that of a single star.

[0295] The above are only preferred embodiments of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for constructing a multi-satellite collaborative strategy for space-based Internet of Things based on an element spatiotemporal expansion graph, characterized in that: The following steps are involved: Step 1: Construct a weighted spatiotemporal connection graph (STCG) to stabilize the satellite dynamic network. Step 2: Based on the weighted spatiotemporal connection graph (STCG), an element weighted spatiotemporal connection graph (ESTCG) is constructed to perform integrated management of device resources, link resources, storage resources, and computing resources in the satellite network. The details are as follows: (1) Nodes: The nodes of the element graph are divided into two categories: satellites and functional modules; Functional modules are abstracted as nodes directly connected to satellites. Since the functional modules equipped by satellite nodes do not change over time, there is no need to consider time slot changes when discussing the feature graph; Assume that the satellite network provides Num types of functional modules, and all the x-th, 1≤x≤Num types of functional modules are represented as an array in Represents satellite v j Therefore, all functional module nodes in the satellite network can use F = {f 1 ,f 2 ,…,f Num Due to the limitations of computing resources and storage capacity, each edge satellite is only equipped with some functional modules, so some elements in the set F are empty. (2) Edges: The edges in the feature graph are divided into two categories: edges between satellites or edges between satellites and functional modules; The edge between the satellite and the functional module indicates that the satellite is equipped with a functional module. The edge from the satellite to the functional module indicates the process of the corresponding functional module processing data, and vice versa indicates the process of returning results. Located in v j The data delay of the functional processing unit data of type x, 1≤x≤Num, that is, 1 bit, is expressed as where c jx (cycles / s) is v j The CPU frequency of the x type function; if v j There is no function of type x, c jx =0(cycles / s) and ct jx =∞(s / bit); therefore, the weight of the edge from the satellite node to the functional module of type x,1≤x≤Num is recorded as The function module is directly installed on the satellite, and the process of returning the result to the satellite no longer consumes time, and the weight of the edge from the function of type x,1≤x≤Num to the satellite node is Except for the edges between satellites, the edges from satellite to functional modules, and the edges from functional modules to satellites, there are no other edges in the element graph within a single time slot, and the weight of disconnected edges is infinite; In k time slots, the weighted adjacency matrix of the satellite network represented by ESTCG is obtained: Where Inf represents a matrix whose diagonal elements are 0 and the rest of the elements are ∞; In adjacent time slots {k, k+1}, the weighted adjacency matrix of the satellite network topology with added elements is shown as The weighted adjacency matrix of the satellite network after combining time slot division with feature graph is represented as Among them, GM N,1 Represents the periodicity of the satellite dynamic topology changes; accordingly, the dynamic topology of the satellite network and the data transmission and processing process in the satellite network are restored according to GraphM; Step 3: Based on the element-weighted spatiotemporal connectivity graph (ESTCG), a multi-satellite collaboration strategy for the space-based Internet of Things is constructed.

2. The method for constructing a multi-satellite collaborative strategy for space-based Internet of Things based on an element spatiotemporal extension graph according to claim 1 is characterized in that: In step 1, a weighted space-time connection graph (STCG) is constructed. The conditions for constructing intersatellite links include the satellite's position, geometric visibility conditions, antenna visibility conditions, and channel capacity requirements, where: (1) Satellite location Based on the WGS-84 coordinate system: the origin coincides with the center of the Earth, the Z axis points to the CTP direction of the Earth's poles as defined by BIH 1984.0, the X axis points to the intersection of the zero meridian plane of BIH 1984.0 and the CTP equator, and the Y axis, Z axis, and X axis form a right-handed coordinate system; The position of the satellite is determined by the six major Kepler parameters. The semi-major axis a and the eccentricity e determine the shape of the satellite orbit. The satellite orbit plane is determined by the inclination δ and the right ascension Ω of the ascending node. The perigee angular distance ω determines the position of the orbit on the orbital plane, and the true anomaly f determines the position of the satellite in the orbit. Among the six Kepler parameters that describe the satellite's undisturbed motion, only the true anomaly is a function of time, while the others are constants. Therefore, the calculation of the satellite's instantaneous position lies in calculating the true anomaly. Assuming that satellite v i The six Kepler parameters are The subscript i represents the satellite number. To calculate the true anomaly, two auxiliary parameters are introduced, namely the eccentric anomaly E and i Peace anomaly M i , where M i The expression of time variation is: Where G is the gravitational constant, M is the mass of the Earth, Satellite v i The time of passing the perigee, t is the time of observing the satellite; according to Kepler's equation, the eccentric anomaly angle E i for E i (t)=M i (t)+e i sin E i (t) (2) Eccentric anomaly E i Solve it by iterative method and get the true perigee f i : Satellite v i The distance to the center of the earth is r i (t)=a i (1-cosE i (t)) (4) Then satellite v i The coordinates in the WGS-84 coordinate system (x i (t),y i (t),z i (t)) is Therefore, the satellite v i With v j The distance between (2) Geometric visibility conditions Assuming the thickness of the atmosphere is h, when the communication link between two satellites is tangent to the atmosphere, this is the critical state of geometric visibility between satellites; Two satellites v i and v j The following conditions must be met to establish an intersatellite link: d ij (t)≥(R e +h) (7) Among them, R e is the radius of the earth; the above formula is converted into a constraint on the satellite distance: (3) Antenna visibility conditions Assuming that the satellite antennas have sufficient tracking capabilities, when two satellites are in each other's antenna beam, they are considered to be visible between the satellite antennas. The satellite antennas are fixed to the ground, and the satellite antenna scanning range is considered to be a cone with the satellite antenna transmitting point as the vertex and the maximum scanning range of the antenna beam as the cone angle. The height of the cone is aligned with the line connecting the satellite antenna transmitting point and the center of the earth. Assume satellite v i The beam scanning range is γ i , satellite v j The beam scanning range is γ j , under the condition that the satellite meets the geometric visibility, the four antenna visibility situations include: satellite v i and v j Both satellites are outside the scanning range of each other's beam, and both satellites are invisible; satellite v i In satellite v j Within the beam scanning range, but the satellite v j Not on Satellite V i Within the beam scanning range, the two satellites are still not visible; Satellite v i and satellite v j Both satellites are just within the scanning range of each other's beams, and the two satellites are just visible; satellite v i and satellite v j Normally visible; When the antenna visibility condition is met, the center of the earth and the satellite v j 、Satellite v i The central angle θ of the circle should satisfy: Similarly, the above expression can be converted into the distance expression: (4) Channel capacity requirements In natural space, satellite v i ,v j , the propagation loss between i≠j is in, is the operating wavelength, f is the carrier frequency, and c is the speed of light; Based on the free space propagation formula, v i ,v j , the power of the received signal between i≠j is as follows: Among them, P e is the transmitting power of the transmitting source, G e is the transmission gain of the directional antenna, G r is the receiving antenna gain, and P e G e It is called the effective isotropic radiated power EIRP, and the signal-to-noise ratio at the receiving end is expressed as Where k = -228.598 dBW / Hz is the Boltzmann constant, T represents the noise temperature, N0 is the noise power spectrum density, and B is the bandwidth of the intersatellite link. According to the Shannon formula, the intersatellite channel capacity is C ij (t)=Blog2(1+S ij (t)) (14) Assume that the minimum communication channel capacity between satellites is C0, that is, C ij (t)>C0, the two satellites v i ,v j , i≠j can communicate at time t, otherwise not; then under the condition of signal power attenuation, satellite v i ,v j , i≠j can establish a satellite link only if Similarly, the above expression can be converted into the distance expression: v i ,v j , the expression of the intersatellite link status between i≠j changing with time is: Among them, g ij (t) = 1 represents satellite v i ,v j ,The link with i≠j is connected at time t, otherwise it is disconnected.

3. The method for constructing a multi-satellite collaborative strategy for space-based Internet of Things based on an element spatiotemporal extension graph according to claim 2 is characterized in that: The stabilization process of the satellite dynamic network is performed as described in step 1, as follows: Based on the discrete topological sequence, time slot partitioning is applied to truncate satellite links with continuously changing states into periodically unchanged satellite link segments. The satellite's operating cycle T is divided into N consecutive and equal time slots, with each time slot lasting Δt = T / N. The satellite network is considered static within each time slot, and the topological structure is viewed as a static steady-state graph that connects multiple satellite nodes according to the existence of satellite inter-satellite links. In order to establish connections between multiple steady-state graphs and describe the global topological information of the satellite time-varying network, virtual links are established between the same satellite nodes in adjacent time slots, thus forming a virtual topology. The steady-state graphs are connected in chronological order to form a complete weighted spatiotemporal connection graph (STCG) of the satellite network throughout the entire cycle. In order to find the routing strategy with the lowest delay in data transmission between satellites, the data transmission delay in the link and the storage delay in the node are used as the weights of the actual link and virtual link, respectively. That is, the data transmission delay in space is used as the weight of the actual link within the time slot, and the data transmission delay in time is used as the weight of the virtual link between time slots. Therefore, the weight of the weighted space-time connection graph STCG is also determined, and the weighted adjacency matrix of the weighted space-time connection graph STCG is written accordingly.

4. The method for constructing a multi-satellite collaborative strategy for space-based Internet of Things based on an element spatiotemporal extension graph according to claim 3 is characterized in that: In step 1, the satellite dynamic network is stabilized, including the construction of satellite topology within the time slot, satellite topology between time slots, and STCG, as follows: (1) Satellite topology within a time slot Assume that the available set of times at the beginning of each time slot is T = {t1, t2, ..., t N } means, where t k ∈T represents the time when the kth time slot starts, and t k+1 -t k =Δt; In order to represent the change of node status over time, suppose there are M nodes in the satellite network and the available set of nodes in the kth time slot is Indicates; v 编号 The form represents the satellite node itself, that is, The form only represents the nodes within a certain time slot; In a single time slot, any two satellites v i ,v j i≠j in the kth time slot, that is, node The average link transmission rate between When transmitting data unit data (1 bit) in a satellite network, the data is transmitted between any two nodes in the kth time slot. The transmission delay between The weighted adjacency matrix of the satellite network topology at the kth time slot is expressed as represents the satellite v i ,v j ,i≠j is not connected in the kth time slot and data transmission is not possible; (2) Satellite topology between time slots Introduce virtual links between different replica nodes of the same satellite in adjacent time slots along the time growth direction, that is, nodes With node It is one-way connected in time; since the virtual link represents the topological connection of the nodes in time, there is no virtual link between the replica nodes of different satellites in different time slots, that is, when i≠j, the node With node Not connected; A virtual link only represents the temporal continuity of the topology. A virtual link means that data is stored in a node and waits for transmission in the next transmission slot. Therefore, the weight of the virtual link should be the remaining time of the current slot when the data is transmitted to the node. Assume that data arrives at the node The delay in the kth time slot is The data is at the node The transmission delay between The weighted adjacency matrix of the satellite network topology structure in adjacent time slots {k, k+1} is expressed as (3) Construction of STCG The weighted adjacency matrix within and between time slots includes the entire process of topological connection, data transmission and storage between satellites. Therefore, all weighted adjacency matrices within the operating cycle [0, T) are combined in the order of time slot changes to record all information transmitted in the satellite dynamic network. The combined matrix is ​​the weighted adjacency matrix of STCG, as shown in the following formula: Among them, G N,1 It represents the periodicity of the dynamic topology changes of the satellite. Accordingly, the dynamic topology of the satellite network and the data transmission process in the satellite network are restored according to the Graph.

5. The method for constructing a multi-satellite collaborative strategy for space-based Internet of Things based on an element spatiotemporal extension graph according to claim 1 is characterized in that: In step 3, a multi-satellite collaboration strategy for the space-based Internet of Things is constructed based on the element-weighted spatiotemporal connectivity graph (ESTCG), as follows: (1) Mission coordination between the two satellites Assume that at time q, A pending service is initiated and a service with a workload of D bits is sent to the satellite v j In the process, the satellite v j The delay in processing this service is expressed as The input values ​​of the function shortest(G, t, v, f) are the weighted adjacency matrix G of the graph, the service start time t, the service starting point v, and the function module f as the end point; Although ESTCG statically represents satellite networks as a weighted adjacency matrix GraphM, compared to quasi-static networks, it still has the following limitations when processing services: First, due to the dynamic nature of satellite networks, the topological connectivity between two time slots may vary. Therefore, the total data transmission delay within a single time slot cannot exceed the time slot Δt, otherwise there is a risk of transmission interruption. Assuming that the weight of each connected link is 0.4Δt, within a time slot, data can only reach node v3 from node v1 and cannot reach node v5. Secondly, the weight of the virtual link in the ESTCG model represents the process of data storage and waiting for transmission; the weight of the virtual link is not fixed. The weight of the virtual link between two time slots of a node is equal to the remaining time of the current time slot when the data arrives at the node; In seeking When the optimal routing is between If the data from the node Transmit to node The meaning of this is that the data is temporarily stored in node v1 for a time slot, so the weight between time slots is is Δt; and the data passes through the time slot The weight of the time slot is 0.4Δt, then the data has spent 0.4Δt in the current time slot when it reaches node v2; When , it means that the data is temporarily stored in node v2 for the remaining time of the kth time slot, that is, the time slot weight is 0.6Δt; The sum of the weights of the shortest routes between them is 1.4Δt; When measuring service processing delay, it is necessary to add restrictions based on formula (30): ① Assume any two satellites and When the amount of data transmitted is D bit, The shortest delay of the optimal route transmission unit delay between For data to be successfully transmitted within a single time slot, the following conditions must be met: ② Since the weight of the cross-time slot edge is not fixed and is related to the time when the data arrives at the node, the weight of the cross-time slot edge needs to be calculated in real time based on the choice of data transmission path; (2) Space-based Internet of Things mission coordination Assume that at time q, A pending service is initiated. The data volume of this task is Dbit and needs to be processed by a function module of type x. Assume that the data volume ratio of the task to each satellite in the satellite network is {λ1,λ2,…,λ M }, that is, assigned to the functional module f x The amount of data is {λ1×D,λ2×D,…,λ M ×D}, the delay of space-based Internet of Things collaborative processing of this service is the delay of the latest task assigned to it in the functional module, and can be expressed as In order to reduce the processing delay of services as much as possible, the goal of multi-satellite collaboration in space-based IoT is to minimize the service processing delay. The optimization problem is then expressed as And the constraints must be met: ① Assume any two satellites and When the amount of data transmitted is Dbit, The shortest delay for transmitting unit data through the best route between For data to be successfully transmitted within a single time slot, the following conditions must be met: ② Since the weight of the cross-time slot edge is not fixed, it is related to the time when the data arrives at the node. Therefore, the weight of the cross-time slot edge needs to be calculated in real time based on the choice of data transmission path; The BPSO algorithm is used to solve this optimization problem. Assuming that the size of the particle swarm is U and I represents the maximum number of iterations of the swarm, the position and velocity of the u≤Uth particle in the i≤Ith iteration are as follows: Where, The value in represents the proportion of data allocated to each satellite, that is, The actual meaning is λ j The value of The value of represents the particle moving step to find the optimal solution, and the value is between [0,1]. When i<I, particle u first tracks the historical optimal position p. ubest and the historical optimal position g of the entire particle swarm best Update speed, as shown below: Where μ is the inertia weight, γ1 and γ2 are acceleration factors, and β1 and β2 are random numbers uniformly distributed in the interval [0,1]. The direction updates the position as follows: symbol Indicates that the value is rounded up, and Represents the permutation operation between the elements of two matrices. The specific process is to first determine The index of the element with the largest value in is taken as the position to be replaced, and then the value is calculated according to the constraints in the above formula. The index of the element that is to be replaced with the position to be replaced is used to complete the replacement operation; finally, the fitness value is calculated by the following formula: In summary, the optimal inter-satellite collaborative computing solution is solved based on the BPSO algorithm.

Images