Enhanced Interference Management for 6G in-X Subnetworks

Abstract

Short-range low-power 6th generation (6G) in-X subnetworks are proposed as a viable radio concept for supporting extreme communication requirements in emerging applications such as wireless control of robotic arms and control of critical on-body devices, e.g. wireless heart pacemaker. For these applications, ultra-high reliability (e.g., above 6 nines) with sub-ms latency must be guaranteed at all spatiotemporal instants. To meet these requirements, radio systems that are robust against fading and interference are crucial. In this paper, we present a comprehensive investigation on technology enablers and techniques for interference mitigation in 6G in-X subnetworks. We present several techniques including blind-repetition with pseudo-random frequency hopping and environment-aware mechanisms for interference management via dynamic channel allocation. We further propose two novel enhancements viz: (1) repetition order adaptation involving real-time selection of the number of repetitions based on current channel conditions; and (2) anticipatory packet duplication in which each subnetwork duplicates its transmission on a secondary channel group whenever it detects the presence of a potentially harmful neighbouring subnetwork. We perform extensive simulations in an industrial factory environment with mobile in-X subnetworks using models and parameters defined by the 3rd Generation Partnership Project (3GPP). Results show that in-X subnetworks requires large bandwidth (≥ 1 GHz), up to 2 packet repetitions and environment-aware interference coordination in order to support packet loss rate below 10 −6 with a latency < 100µs. The number of repetitions can however be reduced for systems with survival time greater than the cycle time. The proposed enhancements also result in up to ×10 4 packet loss rate reduction for systems with survival time above 2 cycle times.

Power Control for 6G In-Factory Subnetworks With Partial Channel Information Using Graph Neural Networks

“…The subnetworks are indexed by n ∈ {1, • • • , N } and devices by j ∈ {1, • • • , J}. The devices within the subnetworks are scheduled by their AP over orthogonal frequency and time slots to avoid intrasubnetwork interference [2], [22], [42]. As shown in Figure 1, we assume that the subnetworks are time synchronized and utilize a common frequency-time resource grid, where a device is scheduled over a single resource unit.…”

Distributed Channel Allocation for Mobile 6G Subnetworks via Multi-Agent Deep Q-Learning

“…The target network has the same structure as the main network but its weights are only updated after a specified number of steps, T update , i.e., θt := θ t every T update steps. Decay exploration probability as in (12).…”

“…Similar to the works in [5], [12], we utilized random switching delays to minimize the impact of ping-pong effects resulting from simultaneous switching by multiple subnetworks to the same channel. The delay is generated for all subnetworks at the beginning of each snapshot as a random integer factor of the transmission interval with a maximum value of 10.…”

“…While several solutions have been proposed for resource allocation in different wireless systems over the years, works targeting the peculiar nature of short-range low-power 6G in-X subnetworks are still rather limited. In our previous works, we have proposed distributed rule-based heuristics [5], [12] and a supervised learning method [13] in which a deep neural network (DNN) is trained with data generated using centralized graph coloring for channel allocation in scenarios with dense deployment of 6G in-X subnetworks. In a recent work [14], a Q-learning method for joint power and channel allocation using quantized state information is proposed.…”

Channel Prediction for Mobile MIMO Wireless Communication Systems

“…We consider deployments with no transmitter-to-receiver distance restriction and with a maximum transmitter-to-receiver distance of d max = 3 m which are henceforth denoted Scen 1 and Scen 2, respectively. While Scen 1 corresponds to a typical D2D communication scenario, Scen 2 represents scenarios where a transmitter communicates with a receiver in close proximity as in, e.g., short-range low-power 6G in-X subnetworks [15], [16] for extreme wireless communication inside entities such as the human body, industrial robots, and vehicles. Transmission bandwidth of B = 5 MHz with carrier frequency, f c = 6GHz is considered in the simulations.…”