In vehicular ad hoc networks (VANETs), emergency messages broadcasting has been considered as one of the important parts for safety related applications. The efficiency and reliability of the emergency message broadcasting are severely affected by the network structure, message redundancy, channel contention, etc. To address this problem, in this paper, we propose an RSUs-assisted hybrid emergency message broadcasting (RA-HEMB) protocol for two-way grid roads in urban VANETs. In doing so, the broadcasting target region is first generated based on the instantaneous traffic status and types of the emergency messages. To balance the deliver latency and the reliability, an adaptive forwarding nodes selection scheme based on position is proposed to dynamically determine the forwarding area according to different vehicle densities and vehicle-to-vehicle (V2V) communication ranges. Then, based on the forwarding nodes selection scheme and roadside unit (RSU) deployment, an RA-HEMB protocol is established by properly utilizing both the wired links via RSUs and the wireless links to broadcast the emergency message in hybrid vehicular networks. Simulation results show that the proposed RA-HEMB can greatly shorten the broadcasting time needed to notice the nodes within the target region performance compared to the benchmark protocols.

Wireless traffic prediction is essential to developing intelligent communication networks that facilitate efficient resource allocation. Along this line, decentralized wireless traffic prediction under the paradigm of federated learning is becoming increasingly significant. Compared to traditional centralized learning, federated learning satisfies network operators’ requirements for sensitive data protection and reduces the consumption of network resources. In this paper, we propose a novel communication-efficient federated learning framework, named FedCE, by developing a gradient compression scheme and an adaptive aggregation strategy for wireless traffic prediction. FedCE achieves gradient compression through top-K sparsification and can largely relieve the communication burdens between local clients and the central server, making it communication-efficient. An adaptive aggregation strategy is designed by quantifying the different contributions of local models to the global model, making FedCE aware of spatial dependencies among various local clients. We validate the effectiveness of FedCE on two real-world datasets. The results demonstrate that FedCE can improve prediction accuracy by approximately 27% with only 20% of communications in the baseline method.

Lysine Malonylation (Kmal) is a newly discovered protein post-translational modifications (PTMs) type, which plays an important role in many biological processes. Therefore, identifying and understanding Kmal sites is very critical in the studies of biology and diseases. The typical methods are time-wasting and expensive. Nowadays, many researchers have proposed machine learning (ML) methods to deal with PTMs’s identification issue. Especially, some deep learning (DL) methods are also utilized in this field. In this work, we proposed K_net, which employed Convolutional Neural Network to identify the potential sites. Meanwhile, we proposed a new verification method Split to Equal Validation (SEV), which can well solve the impact of sample imbalance on prediction results. More Specifically, Acc, Sn, Sp, MCC and AUC values were adopted to evaluate the prediction performance of predictors. In total, CNNKmal achieved the better performance than other methods.