Adapting to Dynamic LEO-B5G Systems: Meta-Critic Learning Based Efficient Resource Scheduling

in IEEE Transactions on Wireless Communications (2022), 21(11), 9582-9595

Low earth orbit (LEO) satellite-assisted communications have been considered as one of the key elements in beyond 5G systems to provide wide coverage and cost-efficient data services. Such dynamic space ... [more ▼]

Low earth orbit (LEO) satellite-assisted communications have been considered as one of the key elements in beyond 5G systems to provide wide coverage and cost-efficient data services. Such dynamic space-terrestrial topologies impose an exponential increase in the degrees of freedom in network management. In this paper, we address two practical issues for an over-loaded LEO-terrestrial system. The first challenge is how to efficiently schedule resources to serve a massive number of connected users, such that more data and users can be delivered/served. The second challenge is how to make the algorithmic solution more resilient in adapting to dynamic wireless environments. We first propose an iterative suboptimal algorithm to provide an offline benchmark. To adapt to unforeseen variations, we propose an enhanced meta-critic learning algorithm (EMCL), where a hybrid neural network for parameterization and the Wolpertinger policy for action mapping are designed in EMCL. The results demonstrate EMCL’s effectiveness and fast-response capabilities in over-loaded systems and in adapting to dynamic environments compare to previous actor-critic and meta-learning methods. [less ▲]

Efficient Resource Scheduling and Optimization for Over-Loaded LEO-Terrestrial Networks

in Proceeding of IEEE ICC 2022 (2022, May)

Towards the next generation networks, low earth orbit (LEO) satellites have been considered as a promising component for beyond 5G networks. In this paper, we study downlink LEO-5G communication systems ... [more ▼]

Towards the next generation networks, low earth orbit (LEO) satellites have been considered as a promising component for beyond 5G networks. In this paper, we study downlink LEO-5G communication systems in a practical scenario, where the integrated LEO-terrestrial system is over-loaded by serving a number of terminals with high-volume traffic requests. Our goal is to optimize resource scheduling such that the amount of undelivered data and the number of unserved terminals can be minimized. Due to the inherent hardness of the formulated quadratic integer programming problem, the optimal algorithm requires unaffordable complexity. To solve the problem, we propose a near-optimal algorithm based on alternating direction method of multipliers (ADMM-HEU), which saves computational time by taking advantage of the distributed ADMM structure, and a low-complexity heuristic algorithm (LC-HEU), which is based on estimation and greedy methods. The results demonstrate the near-optimality of ADMM-HEU and the computational efficiency of LC-HEU compared to the benchmarks. [less ▲]

Sparsification and Optimization for Energy-Efficient Federated Learning in Wireless Edge Networks

in IEEE (2022)

Federated Learning (FL), as an effective decentral- ized approach, has attracted considerable attention in privacy- preserving applications for wireless edge networks. In practice, edge devices are ... [more ▼]

Federated Learning (FL), as an effective decentral- ized approach, has attracted considerable attention in privacy- preserving applications for wireless edge networks. In practice, edge devices are typically limited by energy, memory, and computation capabilities. In addition, the communications be- tween the central server and edge devices are with constrained resources, e.g., power or bandwidth. In this paper, we propose a joint sparsification and optimization scheme to reduce the energy consumption in local training and data transmission. On the one hand, we introduce sparsification, leading to a large number of zero weights in sparse neural networks, to alleviate devices’ computational burden and mitigate the data volume to be uploaded. To handle the non-smoothness incurred by sparsification, we develop an enhanced stochastic gradient descent algorithm to improve the learning performance. On the other hand, we optimize power, bandwidth, and learning parameters to avoid communication congestion and enable an energy-efficient transmission between the central server and edge devices. By collaboratively deploying the above two components, the numerical results show that the overall energy consumption in FL can be significantly reduced, compared to benchmark FL with fully-connected neural networks. [less ▲]

A Deep Learning Based Acceleration of Complex Satellite Resource Management Problem

Poster (2022)

An SDN Based Testbed for Dynamic Network Slicing in Satellite-Terrestrial Networks

in IEEE MeditCom proceeding (2021)

6G networks are expected to meet ambitious perfor- mance parameters of coverage, data rates, latency, etc. To fulfill these objectives, the implementation of non-GEO satellite con- stellations is expected ... [more ▼]

6G networks are expected to meet ambitious perfor- mance parameters of coverage, data rates, latency, etc. To fulfill these objectives, the implementation of non-GEO satellite con- stellations is expected to improve coverage, capacity, resilience, etc. as well as the implementation of new advanced network virtualization algorithms in order to optimize network resources. However, the integration of these technologies represents new challenges, such as the execution of network slicing schemes in highly dynamic environments and network awareness require- ments. In this regard, Software Defined Networking (SDN) is seen as a required 6G technology enabler in order to provide better satellite-terrestrial integration approaches and Virtual Network (VN) implementation solutions. In this paper, we present an experimental testbed for non-GEO satellite constellations integration solution and VNE algorithms implementation adapted to highly variable network conditions that builds upon SDN. A laboratory testbed has been developed and validated, consisting in SDN-based satellite-terrestrial dynamic substrate network emulated in Mininet, a Ryu SDN controller with an End-to-End (E2E) Traffic Engineering (TE) application for the VNs estab- lishment and a Virtual Network Embedding (VNE) algorithm implemented in Matlab. [less ▲]