Search

Federated learning has been widely employed in many applications to protect the data privacy of participating clients. Although the dataset is decentralized among training devices in federated learning, the model parameters are usually stored in a centralized manner. Centralized federated learning is easy to implement; however, a centralized scheme causes a communication bottleneck at the central server, which may significantly slow down the training process. To improve training efficiency, we investigate the decentralized federated learning scheme. The decentralized scheme has become feasible with the rapid development of device-to-device communication techniques under 5G. Nevertheless, the convergence rate of learning models in the decentralized scheme depends on the network topology design. We propose optimizing the topology design to improve training efficiency for decentralized federated learning, which is a non-trivial problem, especially when considering data heterogeneity. In this paper, we first demonstrate the advantage of hypercube topology and present a hypercube graph construction method to reduce data heterogeneity by carefully selecting neighbors of each training device—a process that resembles classic graph embedding. In addition, we propose a heuristic method for generating torus graphs. Moreover, we have explored the communication patterns in hypercube topology and propose a sequential synchronization scheme to reduce communication cost during training. A batch synchronization scheme is presented to fine-tune the communication pattern for hypercube topology. Experiments on real-world datasets show that our proposed graph construction methods can accelerate the training process, and our sequential synchronization scheme can significantly reduce the overall communication traffic during training.

In the electric spring (ES) topology, traditional smart load (SL) consists of non-critical loads (NCL) and ES in series to regulate power and stabilize the voltage of critical load (CL). Aiming at the problem that NCL cannot always obtain stable voltages, this paper derives the circuit diagram of ES and CL in series to form new SL to stabilize voltage of NCL through circuit transformation. The new SL combines with the traditional SL to form switchable SL. The circuit topology of switchable SL is deduced, and the basic principle of switchable SL is analyzed. The control strategy of ES with switchable SL is studied. The simulation experiment results in Matlab/Simulink show that the voltage of different loads can be stabilized by controlling ES with switchable SL in different time periods.

With the help of instantaneous power theory and Matlab/Simulink simulation tools, the principle of electrical spring (ES) steady-state operation modes division was analyzed, the instantaneous power changes of ES in different steady-state operation modes were studied, then the equivalence circuit of ES under different operating conditions were deduced, the problem that how the steady-state operation mode of ES switches to stabilizes the critical load voltage with the different working conditions was analyzed. Simulation experiments verify the correctness of the research results.