FedSem: A Resource Allocation Scheme for Federated Learning Assisted Semantic Communication

TL;DR

This work tackles the joint optimization of energy, delay, and semantic accuracy in a FedSem system that combines Federated Learning with SemCom over OFDMA. It introduces a two-stage optimization framework that decomposes a challenging non-convex problem into tractable subproblems, using epigraph forms, convexification, SCA, and KKT-based solutions to compute optimal CPU frequencies, compression rate, subcarrier allocation, and transmission powers. The algorithm demonstrates convergence and competitive performance, achieving lower total energy and balanced latency compared with baselines, while accommodating SemCom workload and fairness constraints. The results underscore the potential of resource-aware FedSem deployments for efficient, privacy-preserving semantic communication in wireless networks.

Abstract

Semantic communication (SemCom), regarded as the evolution of the traditional Shannon's communication model, stresses the transmission of semantic information instead of the data itself. Federated learning (FL), owing to its distributed learning and privacy-preserving properties, has received attention from both academia and industry. In this paper, we introduce a system that integrates FL and SemCom, which is called FedSem. We have also proposed an optimization problem related to resource allocation for this system. The objective of this problem is to minimize the energy consumption and delay of FL, as well as the transmission energy of SemCom, while maximizing the accuracy of the model trained through FL. The channel access scheme is Orthogonal Frequency-Division Multiple Access (OFDMA). The optimization variables include the binary (0-1) subcarrier allocation indicator, the transmission power of each device on specific subcarriers, the computational frequency of each participating device, and the compression rate for SemCom. To tackle this complex problem, we propose a resource allocation algorithm that decomposes the original problem into more tractable subproblems. By employing convex optimization techniques, we transform the non-convex problem into convex forms, ensuring tractability and solution effectiveness. Our approach includes a detailed analysis of time complexity and convergence, proving the practicality of the algorithm. Numerical experiments validate the effectiveness of our approach, showing superior performance of our algorithm in various scenarios compared to baseline methods. Hence, our solution is useful for enhancing the operational efficiency of FedSem systems, offering significant potential for real-world applications.

Figures (8)

• Figure 1: The overall system model and the optimization problem formulated in this paper.
• Figure 2: The accuracy versus $\rho$.
• Figure 3: The energy and time consumption under different weight parameters $\kappa_1$, $\kappa_2$ and $\kappa_3$. Note that here when each subfigure has a varying weight parameter, the other weight parameters are set as $1$.
• Figure 4: The energy and time consumption under different $P_{n}^{max}$, where $P_{n}^{max}$ is the maximum transmission power. Here we have $\kappa_1=\kappa_2 = \kappa_3=1$. The compression rate $\rho=1$.
• Figure 5: The energy and time consumption under different subcarriers and users. The maximum transmission power $P_{n}^{max}$ is set to $20$ dB. Here we have $\kappa_1=\kappa_2=\kappa_3=1$. The compression rate $\rho = 1$.

Theorems & Definitions (2)

• Theorem 1
• Theorem 2