We present Tesla-Rapture, a gesture recognition system for sparse point clouds generated by mmWave Radars. State of the art gesture recognition models are either too resource consuming or not sufficiently accurate for the integration into real-life scenarios using wearable or constrained equipment such as IoT devices (e.g., Raspberry PI), XR hardware (e.g., HoloLens), or smart-phones. To tackle this issue, we have developed Tesla, a Message Passing Neural Network (MPNN) graph convolution approach for mmWave radar point clouds. The model outperforms the state of the art on three datasets in terms of accuracy while reducing the computational complexity and, hence, the execution time. In particular, the approach, is able to predict a gesture almost 8 times faster than the most accurate competitor. Our performance evaluation in different scenarios (environments, angles, distances) shows that Tesla generalizes well and improves the accuracy up to 20% in challenging scenarios, such as a through-wall setting and sensing at extreme angles. Utilizing Tesla, we develop Tesla-Rapture, a real-time implementation using a mmWave Radar on a Raspberry PI 4 and evaluate its accuracy and time-complexity. We also publish the source code, the trained models, and the implementation of the model for embedded devices.

BACKGROUND: Molecular interaction networks summarize complex biological processes as graphs, whose structure is informative of biological function at multiple scales. Simultaneously, omics technologies measure the variation or activity of genes, proteins, or metabolites across individuals or experimental conditions. Integrating the complementary viewpoints of biological networks and omics data is an important task in bioinformatics, but existing methods treat networks as discrete structures, which are intrinsically difficult to integrate with continuous node features or activity measures. Graph neural networks map graph nodes into a low-dimensional vector space representation, and can be trained to preserve both the local graph structure and the similarity between node features. RESULTS: We studied the representation of transcriptional, protein-protein and genetic interaction networks in E. coli and mouse using graph neural networks. We found that such representations explain a large proportion of variation in gene expression data, and that using gene expression data as node features improves the reconstruction of the graph from the embedding. We further proposed a new end-to-end Graph Feature Auto-Encoder framework for the prediction of node features utilizing the structure of the gene networks, which is trained on the feature prediction task, and showed that it performs better at predicting unobserved node features than regular MultiLayer Perceptrons. When applied to the problem of imputing missing data in single-cell RNAseq data, the Graph Feature Auto-Encoder utilizing our new graph convolution layer called FeatGraphConv outperformed a state-of-the-art imputation method that does not use protein interaction information, showing the benefit of integrating biological networks and omics data with our proposed approach. CONCLUSION: Our proposed Graph Feature Auto-Encoder framework is a powerful approach for integrating and exploiting the close relation between molecular interaction networks and functional genomics data.

Dual Connectivity (DC) has been proposed by Third Generation Partnership Project (3GPP), in order to address the small coverage areas and outage of users and improve the mobility robustness and rate of users in Heterogeneous Networks (HetNets). In the HetNet with DC, each user is assigned a Macro eNode Base Station (MeNB) and a Small eNode Base Station (SeNB) and transmits data to both eNode Base Stations (eNBs), simultaneously. In this paper, we present a power splitting scheme for the HetNet with DC; to maximize the total rate of the users while not exceeding the maximum transmit power of each user. In our proposed power splitting scheme, a Deep Reinforcement Learning (DRL) approach is taken based on the actor-critic model on continuous state-action spaces. Simulation results demonstrate that our power splitting scheme outperforms the baseline approaches in terms of total rate of users and fairness.

In this paper, a novel self-punishment based scheduling algorithm for a cooperative real- time content download application is designed. In the proposed protocol, selfish mobile devices autonomously form cooperative groups. For each formed group, the base station transmits the content to a selected mobile device designated as seed. Then, the seed shares the content with other mobile devices called sinks over device-to-device links. After analyzing the proposed protocol using a repeated game, new self-punishment mechanisms by revocation or by decreasing the bit rate, are proposed. Such self-punishment mechanisms enable the mobile devices in each cooperative group to autonomously punish selfish seeds without requiring any help from other mobile devices outside cooperative group. ‌Based on the proposed self-punishment mechanisms, a fair algorithm is designed to schedule the seeds in each cooperative group. Then, the designed scheduling algorithm is implemented using an Android application that is developed using Java in the Android Development Tool Bundle. The developed Android application does not depend on the operation system of mobile devices. Simulation results demonstrate that, on the average, the proposed protocol improves the energy efficiency of mobile devices to download real-time content of around 42 % compared to a traditional multicast scenario. Moreover, the proposed protocol does not let the energy efficiency of mobile devices degrade more than 11 %, on average, from the optimal solution even when all the mobile devises are selfish. The practical results show that the maximum difference in the run time of a real- time video over real-world smartphone screens is less than 500 milliseconds when the smartphones form a cooperative group using the developed Android application.