Motor Imagery (MI) is an emerging Brain–Computer Interface (BCI) paradigm in which a person imagines a body movement without any physical action. By decoding the scalp-recorded electroencephalography (EEG) signals, BCIs can establish direct communication pathways to control external devices, offering significant potential in prosthetics, rehabilitation, and human–computer interaction. However, existing solutions remain difficult to deploy in practice. (i) Most employ independent, opaque models for each MI task. This fragmented methodology lacks a unified architectural foundation. Consequently, these models are trained in isolation and fail to learn robust representations from diverse datasets, which often results in modest performance. (ii) They primarily adopt fixed sensor deployment, whereas real-world setups vary in electrode number and placement, causing models trained on one configuration to fail under another. (iii) Performance degrades sharply under low-SNR conditions typical of consumer-grade EEG. Together, these limitations hinder the practical adoption of MI-based BCIs.

This poster introduces SynMotion, a novel mmWave-based human motion sensing system addressing the scarcity of training datasets. By synthesizing mmWave signals using existing vision-based human motion datasets, this system overcomes the challenge of collecting and labeling mmWave data, facilitating wider adoption of mmWave technology for applications like activity recognition, skeleton tracking and radar placement recommendation.

Accurate estimation of on-device model training time is increasingly required for emerging learning paradigms on mobile edge devices, such as heterogeneous federated learning (HFL). HFL usually customizes the model architecture according to the different capabilities of mobile edge devices to ensure efficient use of local data from all devices for training. However, due to oversimplification of latency modeling, existing methods rely on a single coefficient to represent computational heterogeneity, resulting in sub-optimal HFL efficiency. We find that existing methods ignore the important impact of runtime optimization of deep learning frameworks, which we call development-chain diversity. Specifically, layers of a model may have different algorithm implementations, and deep learning frameworks often have different strategies for selecting the algorithm they believe is the best based on a range of runtime factors, resulting in different training latencies and invalid predictions from existing methods. In this paper, in addition to considering this diversity to ensure synchronized completion time of model training, we also study how to select the best algorithm each time to reduce the latency of the per-round training, thereby further improving the overall efficiency of federated training. To this end, we propose LATTE, which consists of a novel selector that identifies the best algorithm at runtime based on relative runtime factors. By further integrating it into our training latency model, LATTE provides accurate training time estimation. We develop LATTE as middleware, compatible with different deep learning frameworks. Extensive results show significantly improved training convergence speed and model accuracy compared to state-of-the-art methods.

This paper introduces ASGaze, a novel gaze tracking system using the RGB camera of smartphones. ASGaze improves the accuracy of existing methods and uniquely tracks gaze points on various surfaces, including phone screens, computer displays, and non-electronic surfaces like whiteboards or paper - a situation that is challenging for existing methods. To achieve this, we revisit the 3D geometric eye model, commonly used in high-end commercial trackers, and it has the potential to achieve our goals. To avoid the high cost of commercial solutions, we identify three fundamental issues when processing the eye model with an RGB camera, including how to accurately extract iris boundary that is the meta-information in our design, how to remove ambiguity from iris boundary to gaze point transformation, and how to map gaze points onto the target surface. Furthermore, as we consider deploying ASGaze in real-world applications, two additional challenges should be addressed: how to automatically and accurately annotate the training dataset to reduce manual labor and time costs, and how to accelerate the inference speed of ASGaze on mobile devices to improve user experience. We propose effective techniques to resolve these issues. Our prototype and experiments on three tracking surfaces demonstrate significant performance gains.

Executing deep neural networks (DNNs) on edge artificial intelligence (AI) devices enables various autonomous mobile computing applications. However, the memory budget of edge AI devices restricts the number and complexity of DNNs allowed in such applications. Existing solutions, such as model compression or cloud offloading, reduce the memory footprint of DNN inference at the cost of decreased model accuracy or autonomy. To avoid these drawbacks, we divide DNN into blocks and swap them in and out in order, such that large DNNs can execute within a small memory budget. Nevertheless, naive swapping on edge AI devices induces significant delays due to the redundant memory operations in the DNN development ecosystem for edge AI devices. To this end, we develop SwapNet, an efficient DNN block swapping middleware for edge AI devices. We systematically eliminate the unnecessary memory operations during block swapping while retaining compatible with the deep learning frameworks, GPU backends, and hardware architectures of edge AI devices. We further showcase the utility of SwapNet via a multi-DNN scheduling scheme. Evaluations on eleven DNN inference tasks in three applications demonstrate that SwapNet achieves almost the same latency as the case with sufficient memory even when DNNs demand$2.32\times$$\sim$$5.81\times$memory beyond the available budget. The design of SwapNet also provides novel and feasible insights for deploying large language models (LLMs) on edge AI devices in the future.

Power governing is a critical component of modern mobile devices, reducing heat generation and extending device battery life. A popular technology of power governing is dynamic voltage and frequency scaling (DVFS), which adjusts the operating frequency of a processor to balance its performance and energy consumption. With the emergence of diverse workloads on mobile devices, traditional DVFS methods that do not consider workload characteristics become suboptimal. Recent application-oriented methods propose dedicated and effective DVFS governors for individual application tasks. Since their approach is only tailored to the targeted task, performance drops significantly when other tasks run concurrently, which is however common on today's mobile devices. In this paper, our key insight is that hardware meta-data, widely used in existing DVFS designs, has great potential to enable capable workload awareness and task concurrency adaptability for DVFS, but they are underexplored. We find that workload characteristics can be described in a hyperspace composed of multiple dimensions derived from these metadata to form a novel workload contextual indicator to profile task dynamics and concurrency. On this basis, we propose a meta-state metric to capture this relationship and design a new solution, GearDVFS. We evaluate it for a rich set of application tasks, and it outperforms state-of-the-art methods.