ACCORD: Application Context-aware Cross-layer Optimization and Resource Design for 5G/NextG Machine-centric Applications

TL;DR

ACCORD tackles the unpredictable QoS demands of machine-centric applications in 5G by introducing a context-aware cross-layer optimization framework that uses a deep reinforcement learning (DRL) agent to jointly configure PHY, MAC, RLC, and APP parameters based on real-time application context. Through experiments in a 3GPP-compliant MATLAB 5G environment, ACCORD demonstrates improved spectrum efficiency and consistent latency for event-driven MCA tasks compared with legacy fixed configurations. Key contributions include MCA requirement characterization, a context-aware DRL policy with a defined state-action-reward structure, and comprehensive performance evaluations across stationary and mobile scenarios. The work highlights a practical path toward Open RAN-enabled adaptive networks that can dynamically meet heterogeneous MCA QoS needs in dynamic wireless environments.

Abstract

Recent advancements in AI and edge computing have accelerated the development of machine-centric applications (MCAs), such as smart surveillance systems. In these applications, video cameras and sensors offload inference tasks like license plate recognition and vehicle tracking to remote servers due to local computing and energy constraints. However, legacy network solutions, designed primarily for human-centric applications, struggle to reliably support these MCAs, which demand heterogeneous and fluctuating QoS (due to diverse application inference tasks), further challenged by dynamic wireless network conditions and limited spectrum resources. To tackle these challenges, we propose an Application Context-aware Cross-layer Optimization and Resource Design (ACCORD) framework. This innovative framework anticipates the evolving demands of MCAs in real time, quickly adapting to provide customized QoS and optimal performance, even for the most dynamic and unpredictable MCAs. This also leads to improved network resource management and spectrum utilization. ACCORD operates as a closed feedback-loop system between the application client and network and consists of two key components: (1) Building Application Context: It focuses on understanding the specific context of MCA requirements. Contextual factors include device capabilities, user behavior (e.g., mobility speed), and network channel conditions. (2) Cross-layer Network Parameter Configuration: Utilizing a DRL approach, this component leverages the contextual information to optimize network configuration parameters across various layers, including PHY, MAC, and RLC, as well as the application layer, to meet the desired QoS requirement in real-time. Extensive evaluation with the 3GPP-compliant MATLAB 5G toolbox demonstrates the practicality and effectiveness of our proposed ACCORD framework.

Figures (8)

• Figure 1: ACCORD framework in 5G. The application data (yellow) and network control data (blue) provide the context that is used as input for the deep reinforcement learning (DRL) to generate the configurations for the network (blue) and application (magenta) for machine-type communication.
• Figure 2: Example scenario for estimating the threshold of round-trip time of video frame generation to inference feedback reception from remote server for a machine-centric task such as smart surveillance.
• Figure 3: Impact of legacy 5G configurations on achieving a target latency of 18 ms for 4 UEs at varying distances from the gNB.
• Figure 4: MLP Model used for training the DQN in ACCORD.
• Figure 5: Rewards achieved by the DRL agent for optimizing the network for different network setup scenarios considered.