Discrete Mode Decomposition Meets Shapley Value: Robust Signal Prediction in Tactile Internet

TL;DR

This paper addresses the challenge of ultra-low latency and reliability in Tactile Internet by introducing a predictive framework that combines Discrete Mode Decomposition (DMD) with Shapley Mode Value (SMV) to forecast haptic signals under adverse network conditions.The proposed approach decomposes discrete haptic signals into intrinsic modes, evaluates each mode's contribution via SMV, and uses this mode-aware input to a Transformer-based predictor, achieving high accuracy with dramatically reduced inference times.Key contributions include a discrete, parameter-light DMD framework with an augmented Lagrangian ADMM optimization for mode extraction, a Shapley-based principled attribution (SMV) for mode selection, and extensive experiments showing up to ~98.9% one-sample accuracy and sub-millisecond inference under challenging conditions.The results demonstrate that DMD+SMV enables robust, real-time haptic prediction, potentially relaxing latency and reliability constraints in TI and providing a scalable path toward real-world TI deployments.

Abstract

Tactile Internet (TI) requires ultra-low latency and high reliability to ensure stability and transparency in touch-enabled teleoperation. However, variable delays and packet loss present significant challenges to maintaining immersive haptic communication. To address this, we propose a predictive framework that integrates Discrete Mode Decomposition (DMD) with Shapley Mode Value (SMV) for accurate and timely haptic signal prediction. DMD decomposes haptic signals into interpretable intrinsic modes, while SMV evaluates each mode's contribution to prediction accuracy, which is well-aligned with the goal-oriented semantic communication. Integrating SMV with DMD further accelerates inference, enabling efficient communication and smooth teleoperation even under adverse network conditions. Extensive experiments show that DMD+SMV, combined with a Transformer architecture, outperforms baseline methods significantly. It achieves 98.9% accuracy for 1-sample prediction and 92.5% for 100-sample prediction, as well as extremely low inference latency: 0.056 ms and 2 ms, respectively. These results demonstrate that the proposed framework has strong potential to ease the stringent latency and reliability requirements of TI without compromising performance, highlighting its feasibility for real-world deployment in TI systems.

Figures (6)

• Figure 1: Overview of three-stage mode-centric predictive framework for TI.
• Figure 2: Error evaluation of Transformer in training phase $W$=1.
• Figure 3: Accuracy and inference time evaluation of DMD and DMD+SMV, and baseline prediction for three NN architectures. The changes of accuracy and inference time from W=1 to 100 are provided for each method. Solid lines indicate prediction accuracy (%) based on the left y-axis, while dashed lines represent inference time (seconds) based on the right y-axis. A green line, aligned with the right y-axis, is the latency constraint.
• Figure 4: Inference Accuracy changes over sliding windows with $W=5$ for predicting next 100-samples.
• Figure 5: Accuracy evaluation of Transformer architectures for sliding $W$=5 in the inference phase.

Theorems & Definitions (1)

• Theorem 1: Shapley Mode Value