Hyperdimensional Computing Empowered Federated Foundation Model over Wireless Networks for Metaverse

TL;DR

This work introduces FSL-HDC, a novel integration of Federated Split Learning (FSL) and Hyperdimensional Computing (HDC) to enable efficient, privacy-preserving foundation-model training over wireless edge networks for Metaverse applications. By partitioning models between clients, a fed server, and a main server and leveraging high-dimensional vector operations, the framework achieves low computational load, fast convergence, and robustness to non-IID data, while reducing data transmission. An auxiliary optimization algorithm jointly tunes transmission power and bandwidth to minimize the maximum upload delay, yielding up to 64% reduction in transmission time. Empirical results on MNIST show FSL-HDC attains about 87.5% accuracy (slightly below FL-HDC) but converges 3.733× faster than FSL-NN and remains robust to data heterogeneity, supporting real-time, privacy-aware Metaverse services.

Abstract

The Metaverse, a burgeoning collective virtual space merging augmented reality and persistent virtual worlds, necessitates advanced artificial intelligence (AI) and communication technologies to support immersive and interactive experiences. Federated learning (FL) has emerged as a promising technique for collaboratively training AI models while preserving data privacy. However, FL faces challenges such as high communication overhead and substantial computational demands, particularly for neural network (NN) models. To address these issues, we propose an integrated federated split learning and hyperdimensional computing (FSL-HDC) framework for emerging foundation models. This novel approach reduces communication costs, computation load, and privacy risks, making it particularly suitable for resource-constrained edge devices in the Metaverse, ensuring real-time responsive interactions. Additionally, we introduce an optimization algorithm that concurrently optimizes transmission power and bandwidth to minimize the maximum transmission time among all users to the server. The simulation results based on the MNIST dataset indicate that FSL-HDC achieves an accuracy rate of approximately 87.5%, which is slightly lower than that of FL-HDC. However, FSL-HDC exhibits a significantly faster convergence speed, approximately 3.733x that of FSL-NN, and demonstrates robustness to non-IID data distributions. Moreover, our proposed optimization algorithm can reduce the maximum transmission time by up to 64% compared with the baseline.

Figures (4)

• Figure 1: Overview of HDC model encoding, training, inference, and retraining.
• Figure 2: The architecture of proposed FSL-HDC framework.
• Figure 3: Accuracy vs. Epochs with different data distribution of FSL-HDC and non-IID of two baselines.
• Figure 4: Minimized transmission time under different available spectrum bandwidth resources.