Bandwidth Allocation Problem Research Articles

A hybrid Bi-level management framework for caching and communication in Edge-AI enabled IoT

The proliferation of IoT devices has led to a surge in network traffic, resulting in higher energy usage and response delays. In-network caching has emerged as a viable solution to address this issue. However, caching IoT data faces two key challenges: the transient nature of IoT content and the unknown spatiotemporal content popularity. Additionally, the use of a global view on dynamic IoT networks is problematic due to the high communication overhead involved. To tackle these challenges, this paper presents an adaptive management approach that jointly optimizes caching and communication in IoT networks using a novel bi-level control method called BC3. The approach employs two types of controllers: a global ILP-based optimal controller for long-term timeslots and local learning-based controllers for short-term timeslots. The long-term controller periodically establishes a global cache policy for the network and sends specific cache rules to each edge server. The local controller at each edge server solves the joint problem of bandwidth allocation and cache adaptation using deep reinforcement learning (DRL) technique. The main objective is to minimize energy consumption and system response time with utilizing the global and local observations. Experimental results demonstrate that the proposed approach increases cache hit rate by approximately 12% and uses 11% less energy compared to the other methods. Increasing the cache hit rate can lead to a reduction in about 17% in response time for user requests. Our bi-level control approach offers a promising solution for leveraging the network's global visibility while balancing communication overhead (as energy consumption) against system performance. Additionally, the proposed method has the lowest cache eviction, around 19% lower than the lowest eviction of the other comparison methods. The eviction metric is a metric to evaluate the effectiveness of adaptive caching approach designed for transient data.

Access strategies in mmWave cell-free network: A matching and auction theory based approach

The mmWave cell-free network is viewed as an energy-efficient and emerging architecture for beyond 5G systems. However, many practical issues should be solved for this system before the benefits are enjoyed, whereof feasible implementation of access strategy is one of the most vital. To this end, the access strategy including user scheduling and bandwidth allocation is investigated in this paper. Firstly, to maximize the defined utility function, the joint optimization problem of user scheduling and bandwidth allocation is formulated as a mixed integer nonlinear programming problem, which is intractable to search for an optimal solution in polynomial time. Secondly, a two-stage matching and sealed-auction-based access strategy is proposed to obtain a sub-optimal solution, where UEs with high-quality channel conditions are scheduled in the first-stage matching as much as possible, and the rest unmatched UEs are scheduled in the second-stage matching with the sealed-auction based bandwidth allocation. Thirdly, a simplified full-matching based access strategy with lower complexity is proposed for the high-dynamic scenario caused by the high-speed movement of UEs, and the complexities and convergence of the two proposed strategies are quantitatively analyzed. Finally, the performance of proposed strategies is evaluated through abundant simulations in different scenarios, and the superiorities of the proposed strategies are demonstrated through the comparison with reference strategies.

Client scheduling and bandwidth slicing for multiple federated learning tasks over multiple passive optical networks

Federated Learning (FL) has attracted extensive attention in facilitating emerging edge intelligence applications for its inherent advantages of ensuring data security and privacy. Especially in the edge computing networks connected by Passive Optical Network (PON) system, FL is introduced to enable applications like autonomous driving, intelligent manufacturing, and precision medicine. However, in this system, the FL deployment over PON inevitably faces challenges induced by the conflict between a large volume of model data with the restrict latency limitation and the confined bandwidth resource of PON, especially for the multiple FL tasks. To address these issues, a novel scheme is proposed for tackling the client distribution and bandwidth allocation problems under the scenario of multiple simultaneous FL tasks supported by the multiple interconnected PON systems, which is consisted of the client scheduling and bandwidth slicing processes. To be specific, an easy-to-implement heuristic algorithm is first performed to assign the client numbers to PONs based on iterative method, with which certain operations are repetitively executed to achieve optimal solutions. And then, the serial bandwidth slicing which adapts the traditional policy, i.e., one-task-per-cycle, to the situation with multiple FL tasks, and parallel slicing with the multi-task-per-cycle, are designed for the investigated system. Furthermore, the simulation system is constructed to verify our method. The corresponding results exhibit that, the largest 43.2 % round time reduction is achieved by client scheduling compared to benchmark without the scheduling. Compared to the benchmark with serial slicing, our scheme can achieve a maximum 50.11 % of the round time decrease. It's also validated that, our proposed client scheduling and parallel bandwidth slicing method can improve the learning efficiency by reducing communication delay, especially for the situation with less client number and smaller FL threshold.

Solution of the Simultaneous Routing and Bandwidth Allocation Problem in Energy-Aware Networks Using Augmented Lagrangian-Based Algorithms and Decomposition

We discuss several algorithms for solving a network optimization problem of simultaneous routing and bandwidth allocation in green networks in a decomposed way, based on the augmented Lagrangian. The problem is difficult due to the nonconvexity caused by binary routing variables. The chosen algorithms, which are several versions of the Multiplier Method, including the Alternating Direction Method of Multipliers (ADMM), have been implemented in Python and tested on several networks’ data. We derive theoretical formulations for the inequality constraints of the Bertsekas, Tatjewski and SALA methods, formulated originally for problems with equality constraints. We also introduce some modifications to the Bertsekas and Tatjewski methods, without which they do not work for an MINLP problem. The final comparison of the performance of these algorithms shows a significant advantage of the augmented Lagrangian algorithms, using decomposition for big problems. In our particular case of the simultaneous routing and bandwidth allocation problem, these algorithms seem to be the best choice.

Efficient Approximation Methods for Lexicographic Max-Min Optimization

Lexicographic max-min (LMM) optimization is of considerable importance in many fairness-oriented applications. LMM problems can be reformulated in a way that allows to solve them by applying the standard lexicographic maximization algorithm. However, the reformulation introduces a large number of auxiliary variables and linear constraints, making the process computationally complex. In this paper, two approximation schemes for such a reformulation are presented, resulting in problem size reduction and significant performance gains. Their influence on the quality of the solution is shown in a series of computational experiments concerned with the fair network dimensioning and bandwidth allocation problem.

Connectivity Status of Intuitionistic Fuzzy Graphs and Its Applications

The connectivity is one of the crucial parameters of network used to transport network flow, routing problems and bandwidth allocation problems. The increased connectivity makes a network more stable. In this paper, a new parameter called connectivity status of a vertex is introduced in the intuitionistic fuzzy graph. The related definitions and propositions of connectivity status of a vertex are proposed and investigated in an intuitionistic fuzzy graph. Specifically, connectivity status and status sequence are defined and analysed with various examples. After deleting a vertex, we classify the vertices of an intuitionistic fuzzy graph as connectivity status enhancing vertices, connectivity status neutral vertices and connectivity status reducing vertices because of the change of connectivity status. Finally, we establish two algorithms for these concepts and give an application to illustrate feasibility of algorithms.

Quality of Service Aware Dynamic Bandwidth Allocation for Rate Control in WSN

Different types of data can be generated by Wireless Sensor Networks (WSNs) in both Real-Time (RT) and Non-RT (NRT) scenarios. The combination of these factors, along with the limited bandwidth available, necessitates careful management of these categories in order to reduce congestion. Due to this, a Proficient Rate Control and Fair Bandwidth Allocation (PRC-FBA) method has been created that prioritizes certain types of traffic and creates a virtual queue for them.In PRC-FBA, the Signal-to-Noise and Interference Ratio (SINR) model is applied to the problem of bandwidth allocation in WSN in an effort to find a compromise between equity and performance. Then, a brand-new bandwidth utility factor is defined with regard to equity and effectivenes. The FBA method in PRC-FBA is devoped for only improving throughput, but not considering delay. However, delay is the main factors for trasnmiitng NRT packets. This paper offers a PRC with Quality of Service (QoS) aware Dynamic Bandwidth Allocation (PRC-QDBA) approach for allocating bandwidth while prioritizing packets based on their traffic classes. This model employs a QoS associated dynamic bandwidth allocation strategy which efficiently distributes the unused time slots among the required nodes. The distribution technique is performed based on hierarchical manner utilizing a parent-child association of tree topology. The parent node receives traffic indication maps (TIMs) from the children nodes and adopts them to allocate time slots based on their demamds. If the parent node is unable to allocate the required slots, it creates a TIM that indicating the demands and transfer it to its immediate parent node. This increases the entire performance rate of RT traffic. Furthermore, this model assures the packet forwarding for previously accepted flows by allowing node transmission based on ancestral connection capabilities. Finally, simulation results demonstartes that the suggested model significantly increases the throughput and delay for bandwidth allocation while also enabling QoS support for RT traffic in WSNs.

Semantic-Driven Efficient Service Network Towards Smart Healthcare System in Intelligent Fabric

The advantages of intelligent fabrics enable health service systems more efficient and safe to perceive and transmit, where massive sensitive data related to health and behavior are collected through the multi-source flexible sensor network. However, due to its complex network structure, health service systems face the challenge of low-power data transmission and processing in intelligent fabric space. Recently, Deep learning-powered semantic communication (Deep-SC) has emerged as a promising communication paradigm. To tackle the above problem, we introduce the concept of semantic cognition into the cyber-physical system of health service and present the Deep-SC based network service framework for the healthcare system, where physical layer blocks are merged into the traditional communication system to jointly optimize the transceiver in communication. Since semantic encoding and decoding require additional knowledge, consider the introduction of the semantic knowledge base in the system to formulate the joint optimization problem of service offloading and bandwidth allocation in the proposed network service framework, with the goals of reliable semantic communication and efficient network transmission. Furthermore, an online algorithm that optimally adapts service offloading and subcarrier allocation decisions to the time-varying channel conditions is required. Finally, we use a reinforcement learning-based decision algorithm to solve combinatorial optimization problems, which further reduces the computing complexity. Experimental results show that the proposed framework achieves significant performance improvements in system energy consumption and reliable transmission, compared with traditional communication strategies.

Optimal Resource Allocation for Random Multiple Access Oriented SCMA-V2X Networks

In this paper, an optimal resource allocation scheme is proposed for random multiple access (RMA) oriented sparse code multiple access (SCMA) based vehicle-to-everything (V2X) networks. We consider an uplink RMA oriented SCMA, where vehicle-to-infrastructure (V2I) users and vehicle-to-vehicle (V2V) users are considered to randomly choose codebooks for transmission from a shared codebook pool. To avoid mutual interference between V2I users and V2V users, the total bandwidth is decomposed into two parts for transmission of V2I users and V2V users. A joint user-codebook selection and bandwidth allocation optimization problem is formulated to maximize the sum rate of V2I users while guaranteeing the reliability requirement of V2V users. This problem is a mixed-integer programming (MIP) problem with probabilistic constraint, thus it is impractical to directly solve. To solve the problem, the probabilistic constraint is converted into a non-probabilistic one by approximation. Subsequently, efficient but suboptimal staged algorithms are proposed to solve the joint optimization problem. Firstly, a new decoupled Q-learning based user-codebook selection algorithm (DQL-UCSA) is proposed to find the optimal user-codebook selection relationships, which completely address codebook collision problem. Under the optimal user-codebook selection relationships determined, the joint optimization problem is transformed into a single objective bandwidth allocation problem. Then, we solve the single objective problem using a convex optimization method, which maximizes the sum rate of V2I users. Simulation results show the DQL-UCSA can converge quickly and enable a significant performance improvement by addressing codebook collisions. Besides, the proposed RMA-SCMA with optimal bandwidth allocation (OBA) is significantly superior to conventional schemes in terms of sum rate.

Fuzzy logic based dynamic wavelength and bandwidth allocation for 5G front haul networks

The Quality of Service (QoS) requirements for 5G are very strict in terms of latency of Fifth Generation (5G) traffic. 5G fronthaul networks between the baseband unit and remote radio heads are planned to use Next Generation - Passive Optical Networks 2 (NG-PON2) to carry their traffic. However, these Time Wavelength Division Multiplexing (TWDM) PON based networks also provide service to other traffic such as residential and corporate users. Therefore, it is crucial to provide QoS guarantees to 5G traffic. The main issue includes allocating transmission over multiple wavelengths while maintaining QoS. Although using all the available wavelengths all the time may seem to be the simple solution, each active wavelength increases the cost of operation. To solve these issues, this paper proposes a technique using fuzzy logic to dynamically allocate wavelength and bandwidth for 5G fronthaul networks. Two types of traffic, i.e., 5G traffic and normal traffic are considered while the problem of Dynamic Wavelength and Bandwidth Allocation (DWBA) is formulated as a queueing theory problem using penalties and a cost function. Penalties are assigned for waiting packets as well as each active wavelength. Then, modeling the number of active wavelengths, the number of packets in the system, the total cost, and the traffic intensity as input fuzzy variables while modeling the change in the number of active wavelengths as the fuzzy output variable. The Mamdani implication is used for the fuzzy inference engine and the height method is used for defuzzification of the output variable. Simulations in MATLAB show that the proposed technique can maintain the latency of 5G traffic lower than the defined threshold while also significantly reducing the number of active wavelengths. A comparison with the existing state-of-the-art techniques shows that the proposed technique results in an improvement of at least 46% and 29% in terms of the average number of active wavelengths and the average latency for 5G traffic, respectively.

Task-Oriented Delay-Aware Multi-Tier Computing in Cell-Free Massive MIMO Systems

Multi-tier computing can enhance the task computation by multi-tier computing nodes. In this paper, we propose a cell-free massive multiple-input multiple-output (MIMO) aided computing system by deploying multi-tier computing nodes to improve the computation performance. At first, we investigate the computational latency and the total energy consumption for task computation, regarded as total cost. Then, we formulate a total cost minimization problem to design the bandwidth allocation and task allocation, while considering realistic heterogenous delay requirements of the computational tasks. Due to the binary task allocation variable, the formulated optimization problem is non-convex. Therefore, we solve the bandwidth allocation and task allocation problem by decoupling the original optimization problem into bandwidth allocation and task allocation subproblems. As the bandwidth allocation problem is a convex optimization problem, we first determine the bandwidth allocation for given task allocation strategy, followed by conceiving the traditional convex optimization strategy to obtain the bandwidth allocation solution. Based on the asymptotic property of received signal-to-interference-plus-noise ratio (SINR) under the cell-free massive MIMO setting and bandwidth allocation solution, we formulate a dual problem to solve the task allocation subproblem by relaxing the binary constraint with Lagrange partial relaxation for heterogenous task delay requirements. At last, simulation results are provided to demonstrate that our proposed task offloading scheme performs better than the benchmark schemes, where the minimum-cost optimal offloading strategy for heterogeneous delay requirements of the computational tasks may be controlled by the asymptotic property of the received SINR in our proposed cell-free massive MIMO-aided multi-tier computing systems.

Joint user association and caching placement for cache-enabling UAV networks

Cache-enabling unmanned aerial vehicles (UAVs) are considered for storing popular contents and providing downlink data offloading in cellular networks. In this context, we formulate a joint optimization problem of user association, caching placement, and backhaul bandwidth allocation for minimizing content acquisition delay with consideration of UAVs' energy constraint. We decompose the formulated problem into two subproblems: i) user association and caching placement and ii) backhaul bandwidth allocation. We first obtain the optimal bandwidth allocation with given user association and caching placement by the Lagrangian multiplier approach. After that, embedding the backhaul bandwidth allocation algorithm, we solve the user association and caching placement problem by a three-dimensional (3D) matching method. Then we decompose it into two two-dimensional (2D) matching problems and develop low-complexity algorithms. The proposed scheme converges and exhibits a low computational complexity. Simulation results demonstrate that the proposed cache-enabling UAV framework outperforms the conventional UAV-assisted cellular networks in terms of content acquisition delay and the proposed scheme achieves significantly lower content acquisition delay compared with other two benchmark schemes.

Joint dynamic routing and resource allocation in satellite-terrestrial integrated networks

The Satellite-Terrestrial Integrated Networks (STINs) is considered as a reliable and agile next-generation communication network scheme, because it is flexible to deploy and robust to disasters (e.g., earthquakes, floods, and volcanic eruptions). The routing and resource allocation directly affect the power consumption of satellites in the STINs. However, most of existing research works study these two issues separately, and neglect the impact of path selection on the satellite power consumption. This paper investigates the joint optimization problem of routing, bandwidth allocation, user association, channel allocation, and power allocation of the STINs. We aim at minimizing satellites’ power consumption while satisfying user's Quality of Service (QoS). Because the power consumption of satellites for routing in the space segment is involved with the data rate through satellite gateways in the user segment, this optimization problem is a Multi-Objective Optimization (MOO) problem. To this end, we convert the MOO problem to a Single-Objective Optimization (SOO) problem by linear weighted method. Considering that the SOO problem is a Mixed Integer Nonlinear Programming (MINLP) problem, we decompose it into the resource allocation sub-problem in the user segment and the routing optimization sub-problem in the space segment. And we solve these two sub-problems iteratively. Specifically, we convert the first sub-problem into a convex optimization problem and obtain the current user association and channel allocation. Then we formulate the second sub-problem into a multi-commodity flow problem and solve it based on the obtained user association and channel allocation. The SOO problem is solved by iteratively optimizing these two sub-problems. Experimental results demonstrate that our proposed algorithm is able to get an approximate optimal solution quickly and effectively reduce the power consumption of satellites compared with benchmark algorithms in the STINs.

Semi-Synchronous Personalized Federated Learning Over Mobile Edge Networks

Personalized Federated Learning (PFL) is a new Federated Learning (FL) approach to address the heterogeneity issue of the datasets generated by distributed user equipments (UEs). However, most existing PFL implementations rely on synchronous training to ensure good convergence performances, which may lead to a serious straggler problem, where the training time is heavily prolonged by the slowest UE. To address this issue, we propose a semi-synchronous PFL algorithm, termed as Semi-Synchronous Personalized FederatedAveraging (PerFedS2), over mobile edge networks. By jointly optimizing the wireless bandwidth allocation and UE scheduling policy, it not only mitigates the straggler problem but also provides convergent training loss guarantees. We derive an upper bound of the convergence rate of PerFedS2 in terms of the number of participants per global round and the number of rounds. On this basis, the bandwidth allocation problem can be solved using analytical solutions and the UE scheduling policy can be obtained by a greedy algorithm. Experimental results verify the effectiveness of PerFedS2 in saving the training time as well as guaranteeing the convergence of training loss, in contrast to synchronous and asynchronous PFL algorithms.

Minimizing age of information in the uplink multi-user networks via dynamic bandwidth allocation

Improving the information freshness is critical for the monitoring and controlling applications in the cellular Internet of Things (IoT). In this paper, we are interested in optimizing the bandwidth allocation dynamically to improve the information freshness of the short packet based uplink status updates, which is characterized by a recently proposed metric, age of information (AoI). We first design a status update scheme with channel distribution information (CDI). By relaxing the hard bandwidth constraint and introducing a Lagrangian multiplier, we first decouple the multi-MTCD bandwidth allocation problem into a single MTCD Markov decision process (MDP). Under the MDP framework, after variable substitution, we obtain the single-MTCD status update scheme by solving a linear programming problem. Then, we adjust the Lagrangian multiplier to make the obtained scheme satisfy the relaxed bandwidth constraint. Finally, a greedy policy is built on the proposed scheme to adjust the bandwidth allocation in each slot to satisfy the hard bandwidth constraint. In the unknown environment without CDI, we further design a bandwidth allocation scheme which only maximizes the expected sum AoI drop within each time slot. Simulation results show that in terms of AoI, the proposed schemes outperform the benchmark schemes.

Hierarchical Bandwidth Allocation for Social Community-Oriented Multicast in Space-Air-Ground Integrated Networks

With the rapid advance of wireless communication technologies, the promising space-air-ground integrated networks (SAGINs) are advocated to provide ubiquitous multicast transmission services for the social community constituted by a group of mobile users that have strong social ties and similar content interests. However, due to the limited yet valuable spectrum resources, the network heterogeneity, and diverse service demands of mobile users, it is challenging to efficiently allocate bandwidth for social communities with the objective of achieving satisfactory quality of experience (QoE) in SAGINs. To address this problem, in this paper, we propose a hierarchical bandwidth allocation scheme to enable high-quality multicast services for social communities in SAGINs. Specifically, we first develop a hierarchical bandwidth allocation framework. Wherein, the low earth orbit (LEO) satellite is utilized to provide space-to-air (S2A) unicast bandwidth for unmanned aerial vehicles (UAVs) at a certain price. Each UAV is employed to provide air-to-ground (A2G) multicast bandwidth for ground social communities with a certain A2G multicast bandwidth charge. We then formulate the hierarchical bandwidth allocation problem as a four-stage Stackelberg game, where the target of each participant is to maximize its own utility. Afterward, through the game analysis by the backward induction method, the existence of the Stackelberg equilibrium is proved, where the closed-form solutions on the optimal policies of both the social communities and UAVs are derived by the convex optimization method, and the optimal pricing policies of the LEO satellite is achieved by a proposed gradient descent iteration algorithm. Finally, extensive experiments are conducted to demonstrate that the proposed scheme can greatly increase the utilities of social communities while consuming a less bandwidth compared to conventional schemes.

An adaptive heuristic algorithm to solve the network slicing resource management problem

SummaryArtificial intelligence‐based (AI‐based) network slicing bandwidth allocation enables the 5G/6G service providers to create multiple virtual networks atop a shared physical infrastructure while fulfilling varying end‐user demands. Some researchers argue that AI‐enable network may run the danger of having private information compromised. We still need a backup rule‐based methodology to allocate bandwidth resource to each slice, if the AI‐based method suddenly encounters security issues. To design such a rule‐based methodology, this study attempts to answer two questions: (1) Is the network slicing bandwidth allocation problem the nondeterministic polynomial‐time completeness (NP‐completeness)? (2) Is there a heuristic methodology without any training process, which has equivalent performance compared to the AI‐based methodology? This study first proves the classical network slicing bandwidth allocation problems is NP‐completeness. This shows that the designed heuristic method is inescapably suboptimal to the network slicing bandwidth allocation problem. Secondly, this study proposes the Adaptive Hungarian Algorithm (AHA), which outperforms previous AI‐empowered method and does not need any training process. The experiments demonstrate that AHA reached 93%–97% of the maximal system throughput by brute‐and‐force algorithm, compared to other methodologies only having at most 93% of the maximal system throughput. This also indicates that AHA is capable to solve the network slicing bandwidth allocation problem, if the telecommunication operators do not have sufficient sample complexity to train an AI model.

S-Iteration inertial subgradient extragradient method for variational inequality and fixed point problems

In developing iterative methods for approximating solutions of optimization problems, one of the major goals of researchers is to construct efficient iterative schemes. Over the years, different techniques have been devised by authors to achieve this goal. In this paper, we study a non-Lipschitz pseudomonotone variational inequality problem with common fixed points constraint of Bregman quasi-nonexpansive mappings. We introduce a new iterative method for approximating the solution of this problem in a more general framework of reflexive Banach spaces. Our method employs several techniques to achieve high level of efficiency. One of the techniques employed is the S-iteration process, a method known to be highly efficient in comparison with several of the well-known methods. Moreover, our proposed algorithm utilizes the inertial technique and a non-monotonic self-adaptive step size to guarantee high rate of convergence and easy implementation. Unlike several of the existing results on variational inequality problem with non-Lipschitz operator, the design of our method does not involve any linesearch technique. We obtain strong convergence result for the proposed algorithm without the sequentially weakly continuity condition often assumed by authors to guarantee convergence when solving pseudomonotone variational inequality problems. Furthermore, we apply our result to study utility-based bandwidth allocation problem and image restoration problem. Finally, we present several numerical experiments to demonstrate the efficiency of our proposed method in comparison with existing state-of-the-art methods. Our result extends and improves several of the recently announced results in this direction in the literature.

Two Time-Scale Energy and Spectrum Allocation for MEC Networks with Hybrid Energy Supplies

Integrating power grid and renewable energy to charge mobile-edge computing (MEC) server is a promising solution for green networks, which can reduce the energy consumption of power grid and carbon emission. Intelligent management of energy and bandwidth resources in the two time-scale framework can increase the energy efficiency and network revenue in MEC networks with heterogeneous energy supplies and multiple mobile users (MUs). In this paper, we investigate the joint energy management (for workloads offloading and execution) and bandwidth allocation (for workloads transmission) problem that maximizes the network revenue in the MEC network with hybrid energy supplies. We formulate it into a challenging nonlinear optimization problem because of the data randomness and the temporal coupling effect. Based on the Lyapunov optimization approach with quick convergence speed and low complexity, we propose an online algorithm to overcome the obstacles by relaxing the battery constraints, allowing us to achieve close-to-optimal maximum network revenue with the stability of battery level and different throughput requirements of MUs. Extensive numeral simulations verify the theoretical analysis and demonstrate the superior performance of the proposed algorithm in different algorithms and scenarios.

DNNs Based Computation Offloading for LEO Satellite Edge Computing

Huge low earth orbit (LEO) satellite networks can achieve global coverage with low latency. In addition, mobile edge computing (MEC) servers can be mounted on LEO satellites to provide computing offloading services for users in remote areas. A multi-user multi-task system model is modeled and the problem of user’s offloading decisions and bandwidth allocation is formulated as a mixed integer programming problem to minimize the system utility function expressed as the weighted sum of the system energy consumption and delay. However, it cannot be effectively solved by general optimizations. Thus, a deep learning-based offloading algorithm for LEO satellite edge computing networks is proposed to generate offloading decisions through multiple parallel deep neural networks (DNNs) and store the newly generated optimal offloading decisions in memory to improve all DNNs to obtain near-optimal offloading decisions. Moreover, the optimal bandwidth allocation scheme of the system is theoretically derived for the user’s bandwidth allocation problem. The simulation results show that the proposed algorithm can achieve a good convergence effect within a small number of training steps, and obtain the optimal system utility function values compared with the comparative algorithms under different system parameters, and the time cost of the system and DNNs is very satisfactory.