Recently, decentralized learning has been proposed for model training among mobile devices without center nodes. However, large resource overhead for model aggregation and synchronization would be incurred, which may reduce the learning performance under a given resource budget. To cope with these issues, we propose a novel cluster-based collaborative learning framework over device-to-device (D2D) network, where one device is selected as the cluster head for model aggregation. Within the proposed framework, the learning performance (evaluated by model divergence) and learning latency are analyzed with the consideration of imbalanced data and unreliable D2D communication. Then, an optimization problem is formulated to maximize the learning performance under a given latency constraint by joint cluster head selection and resource allocation. To solve this problem, a lower bound on latency constraint is first obtained for error-free model aggregation. The optimal learning performance is also derived with different degrees of data distribution. After that, an adaptive cluster head selection and resource allocation algorithm is developed for erroneous case by introducing the outage probability. Finally, comprehensive experiments are conducted on well-known models and datasets to illustrate the effectiveness of the proposed algorithm. The results show that our proposal can improve the learning performance while reducing communication and signaling overheads.

The deployment of generative adversarial networks (GANs) in wireless networks faces three key challenges of limited devices’ computational capability, scarce communication resources, and severe data privacy leakage. To address these issues, this paper proposes a new distributed framework for training GANs based on ensemble learning. First, multiple discriminators are trained at many devices using their local datasets. A generator is then trained at a central server by aggregating devices’ discriminators in an ensemble manner. The per-round training time is established. Finally, simulation results show that the proposed framework can simultaneously reduce the training time and improve the learning performance as compared with an existing framework.

In recent years, wirelessfederated learning(FL) has been proposed to support the mobile intelligent applications over the wireless network, which protects the data privacy and security by exchanging the parameter between mobile devices and thebase station(BS). However, the learning latency increases with the neural network scale due to the limited local computing power and communication bandwidth. To tackle this issue, we introduce model pruning for wireless FL to reduce the neural network scale. Device selection is also considered to further improve the learning performance. By removing the stragglers with low computing power or bad channel condition, the model aggregation loss caused by model pruning can be alleviated and the communication overhead can be effectively reduced. We analyze the convergence rate and learning latency of the proposed model pruning method and formulate an optimization problem to maximize the convergence rate under the given learning latency budget via jointly optimizing the pruning ratio, device selection, and wireless resource allocation. By solving the problem, the closed-form solutions of pruning ratio and wireless resource allocation are derived and the threshold-based device selection strategy is developed. Finally, extensive experiments are carried out to demonstrate that the proposed model pruning algorithm outperforms other existing schemes.

Deploying device-to-device (D2D) communication over underlaying cellular network can significantly enhance the spectrum utilization. However, co-channel interference will occur when D2D pairs share the same radio resource with cellular users. To mitigate the interference within a reasonable range, a machine learning based resource reuse scheme for D2D and cellular users is proposed in this paper. Specifically, we formulate an optimization problem to maximize the total throughput of D2D pairs and cellular users by optimally allocating subcarrier and power within the limits of the interference threshold. Since the formulated problem is a mixed integer non-linear programming problem, we solve it in two steps. First, we assign the orthogonal subcarriers to different cellular users to maximize the total throughput of all cellular users. Then, D2D pairs are allowed to reuse different subcarriers to further enhance the throughput without affecting the performance of cellular users. The second step is still NP-hard and therefore we propose a low-complexity algorithm based on the pointer network, a specific neural network structure proposed recently. Results reveal that, with remarkably reduced complexity, the proposed scheme outperforms the conventional resource allocation algorithms.