Dr Yijing Ren

Fog computing is seen as a key enabler to meet the stringent requirements of industrial Internet of Things (IIoT). Specifically, lower latency and IIoT devices’ energy consumption can be achieved by offloading computation-intensive tasks to fog access points (F-APs). However, traditional computation offloading optimization methods often possess high complexity, making them inapplicable in practical IIoT. To overcome this issue, this article proposes a deep reinforcement learning (DRL) based approach to minimize long-term system energy consumption in a computation offloading scenario with multiple IIoT devices and multiple F-APs. The proposal features a multiagent setting to deal with the curse of dimensionality of the action space by creating a DRL model for each IIoT device, which identifies its serving F-AP based on network and device states. After F-AP selection is finished, a low complexity greedy algorithm is executed at each F-AP under a computation capability constraint to determine which offloading requests are further forwarded to the cloud. By conducting offline training in the cloud and then making decisions online, iterative online optimization procedures are avoided and, hence, F-APs can quickly adjust F-AP selection for each device with trained DRL models. Via simulation, the impact of batch size on system performance is demonstrated and the proposed DRL-based approach shows competitive performance compared to various baselines including exhaustive search and genetic algorithm based approaches. In addition, the generalization capability of the proposal is verified as well.

The emerging beyond fifth-generation (B5G) and envisioned sixth-generation (6G) wireless networks are considered as key enablers in supporting a diversified set of applications for industrial mobile robots (MRs). The scenario under investigation in this paper relates to mobile robots that autonomously roam on an industrial floor and perform a variety of tasks at different locations, whilst utilizing high directivity beamformers in millimeter wave (mmWave) small cells. In such scenarios, the potential close proximity of mobile robots connected to different base stations may cause excessive levels of interference having as a net result a decrease in the overall achievable data rate in the network. To resolve this issue, a novel MR optimal path planning scheme via a mixed integer programming formulation is proposed where robots' trajectory is considered jointly with the interference level at different beam sectors. To combat the curse of dimensionality, a geographical division clustering based MR path planning heuristic scheme is proposed to enable scalability and real-time decision making. The proposed heuristic aims to find a low interference path for each mobile robot whilst achieving a near-optimal performance. A wide set of numerical investigations reveal that the proposed optimal and heuristic path planning schemes for the mmWave connected mobile robots can improve the overall achievable throughput by up to 93% compared to an interference oblivious scheme without penalizing the total travel time.

While Vision-Language Models (VLMs) have achieved remarkable performance across diverse downstream tasks, recent studies have shown that they can inherit social biases from the training data and further propagate them into downstream applications. To address this issue, various debiasing approaches have been proposed, yet most of them aim to improve fairness without having a theoretical guarantee that the utility of the model is preserved. In this paper, we introduce a debiasing method that yields a closed-form solution in the cross-modal space, achieving Paretooptimal fairness with bounded utility losses. Our method is training-free, requires no annotated data, and can jointly debias both visual and textual modalities across downstream tasks. Extensive experiments show that our method outperforms existing methods in debiasing VLMs across diverse fairness metrics and datasets for both group and intersectional fairness in downstream tasks such as zero-shot image classification, text-to-image retrieval, and text-toimage generation while preserving task performance.

To address the explosive demand for robust and reliable high-speed communications, millimetre-wave networks that utilize abundant bandwidth in the emerging sixth generation (6G) have been considered as key enablers in supporting diverse applications of industrial mobile robots. Millimeter-wave (mmWave) transmissions are highly susceptible to obstruction by physical obstacles, particularly in cluttered industrial/warehouse environments where mobile robots (MRs) such as humanoids autonomously navigate to carry out diverse tasks. Ensuring robust communication under non-line-of-sight (NLoS) conditions thus remains a significant and largely unresolved challenge. To this end, a novel relay selection based path planning scheme via a mixed integer linear programming formulation is proposed, where two-hop wireless links are established by selecting idle MRs as relay nodes and robots’ trajectory is jointly considered to ensure line-of-sight (LoS) communication. Extensive numerical investigations demonstrate that, compared with a shortest-path scheme that ignores NLoS propagation, the proposed relay-selection-based path-planning scheme improves the overall communication rate by 38%. Moreover, for MRs operating under NLoS conditions, the communication rate can be increased by up to a factor of six.