Communication plays a vital role in multi-agent systems, fostering collaboration and coordination. However, in real-world scenarios where communication is bandwidth-limited, existing multi-agent reinforcement learning (MARL) algorithms often provide agents with a binary choice: either transmitting a fixed amount of data or no information at all. This rigid communication strategy hinders the ability to effectively utilize bandwidth. To overcome this challenge, we present the Dynamic Size Message Scheduling (DSMS) method, which introduces finer-grained communication scheduling by considering the actual size of the information being exchanged. Our approach lies in adapting message sizes using Fourier transform-based compression techniques with clipping, enabling agents to tailor their messages to match the allocated bandwidth according to importance weights. This method realizes a balance between information loss and bandwidth utilization. Receiving agents reliably decompress the messages using the inverse Fourier transform. We evaluate DSMS in cooperative tasks where the agent has partial observability. Experimental results demonstrate that DSMS significantly improves performance by optimizing the utilization of bandwidth and effectively balancing information importance.

Heterogeneities have emerged as a critical challenge in Federated Learning (FL). In this paper, we identify the cause of FL performance degradation due to heterogeneous issues: the local communicated parameters have feature mismatches and feature representation range mismatches, resulting in ineffective global model generalization. To address it, Heterogeneous mitigating FL is proposed to improve the generalization of the global model with resource-independence aggregation. Instead of linking local model contributions to its occupied resources, we look for contributing parameters directly in each node's training results.

Federated Learning (FL) is a popular method for privacy-preserving machine learning on edge devices. However, the heterogeneity of edge devices, including differences in system architecture, data, and co-running applications, can significantly impact the energy efficiency of FL. To address these issues, we propose an energy-efficient personalized federated search framework. This framework has three key components. Firstly, we search for partial models with high inference efficiency to reduce training energy consumption and the occurrence of stragglers in each round. Secondly, we build lightweight search controllers that control the model sampling and respond to runtime variances, mitigating new straggler issues caused by co-running applications. Finally, we design an adaptive search update strategy based on graph aggregation to improve personalized training convergence. Our framework reduces the energy consumption of the training process by lowering the training overhead of each round and speeding up the training convergence rate. Experimental results show that our approach achieves up to 5.02% accuracy and 3.45× energy efficiency improvements.

Federated learning (FL), a new distributed technology, allows us to train the global model on the edge and embedded devices without local data sharing. However, due to the wide distribution of different types of devices, FL faces severe heterogeneity issues. The accuracy and efficiency of FL deployment at the edge are severely impacted by heterogeneous data and heterogeneous systems. In this paper, we perform joint FL model personalization for heterogeneous systems and heterogeneous data to address the challenges posed by heterogeneities. We begin by using model inference efficiency as a starting point to personalize network scale on each node. Furthermore, it can be used to guide the efficient FL training process, which can help to ease the problem of straggler devices and improve FL's energy efficiency. During FL training, federated search is then used to acquire highly accurate personalized network structures. By taking into account the unique characteristics of FL deployment at edge devices, the personalized network structures obtained by our federated search framework with a lightweight search controller can achieve competitive accuracy with state-of-the-art (SOTA) methods, while reducing inference and training energy consumption by up to 3.57× and 1.82×, respectively.

Federated learning, as a distributed machine learning framework that a shared global model that is obtained through frequent local training parameter interaction on each participated device. However, the limited communication bandwidth of participating IoT and edge devices will have a conflict between the frequent parameter-interaction learning mode of federated learning and impact communication and learning efficiency. In this paper, a communication efficiency enhanced federated learning technique is presented by proposing a cooperative filter selection method. The Geometric Median of each layer in the global model is adopted as the criterion to cooperatively select important filters in the local model, and then the corresponding parameters interact with other nodes to achieve efficient communication. Experimental results show that our method has a maximum of $2.66\times$ improvements in communication efficiency compared with the state-of-the-art methods.

Federated learning (FL) is an emerging distributed machine learning technique. However, when dealing with heterogeneous data, a shared global model cannot generalise all devices' local data. Furthermore, the FL training process necessitates frequent parameter communication, which interferes with the limited bandwidth and unstable connections of participating devices. These two issues have a significant impact on FL's effectiveness and efficiency. In this paper, an enhanced communication-efficient personalised FL technique, FedGB, is proposed. Different from existing approaches, FedGB believes that only interacting common information from training results on different devices can improve local personalised training results more effectively. FedGB dynamically selects the backbone structures in the local models to represent the dynamically determined backbone information (common features) in the global model for aggregation. Only interacting common features between different nodes reduce the impact of heterogeneous data to a certain extent. The dynamic adaptive sub-model selection avoids the impact of manually setting the scale of sub-model. FedGB can thus reduce communication overheads while maintaining inference accuracy. The results obtained in a variety of experimental settings show that FedGB can effectively improve communication efficiency and inference accuracy.

The multi-objective neural architecture search (NAS) can automatically realize the network design for high accuracy and high hardware performance for varied applications. However, the existing methods usually need to sample a large number of networks in the search process to guide the controller's search behavior. As a result, the entire search process requires a huge search time overhead. We propose a NAS framework that can perform dynamic adaptive network sampling that regulated by the latency requirements for specific devices and application scenarios. In the layer-wise network sampling process, the sampling probability for each layer is adjusted dynamically according to the current remaining available inference latency space. So that the latency of the sampled network will close to the required latency. Thereby, reducing useless network sampling and improving the search efficiency. Experimental results show that the search efficiency has a 2.35× speedup.

Intelligent unmanned systems (IUSs) are distributed systems composed of multiple agents that share information or cooperate to accomplish specific complex tasks. Agents of the IUS are capable of perception, cognition, control, decision-making, and action. In some cases, the environmental situation and task objectives faced by the IUSs are constantly changing with time. Thus, IUSs are time-sensitive systems. To accelerate the task execution time and response speed, IUSs use artificial intelligence technology to increase the speed and quality of the `observation-orientation-decision-action' (OODA) cycle of task execution. IUSs will tend to decompose the system into different functional units in the future, and individuals take different task roles from the functional perspective of OODA. The system is evolving from a linear OODA cycle of individuals to a cooperative OODA (Co-OODA) with different node roles. At present, the reinforcement learning (RL) algorithm is the mainstream method to solve IUSs cooperation problems. However, it does not adapt to the Co-OODA with different roles; and cannot maximize the Co-OODA system's potential. This paper introduces the role-based Co-OODA system. Furthermore, we propose and design a role-based deep reinforcement learning framework and its corresponding information sharing mechanism.

New generation airborne embedded system has deployed Graphical Processing Units (GPUs) to raise processing capability to meet growing computational demands. Applications in the airborne embedded system have strict real-time constraints. Therefore, it is necessary to accurately predict timing behaviors of those applications. Many previous work propose GPU performance models to estimate the execution time of applications. However, most of those models do not consider the impact of co-execution on the GPU performance. In this paper, we propose a workload-aware GPU performance model to predict the execution time of applications executed concurrently on a single GPU. Experimental results illustrate that the proposed model can achieve a 5.1%-11.6% prediction error in a real airborne embedded hardware platform.

Link prediction is to calculate the probability of a potential link between a pair of unlinked nodes in the future. It has significance value in both theoretical and practical. The similarity of two nodes in the networks is an essential factor to determine the probability of a potential link between them. One of the important methods with the similarity of two nodes is to consider common neighbors of two nodes. However, the number of common neighbors only describes a kind of quantitative relationship without taking into account the topology of given networks and the information of local structure which consist of a pair of nodes and their common neighbors. Therefore, we introduce the concept of the degrees of nodes and the idea of community structure and propose a new similarity index, namely, local affinity structure(LAS). The LAS method describes the closeness of a pair of nodes and their common neighbors. We evaluated LAS on twelve different networks compared with other three similarity based indexes which consider the degree of nodes. From the experimental results, our method shows obvious superiority in improving the accuracy of link prediction.