Motor imagery (MI) electroencephalography (EEG) signals are widely used in BCI systems. MI tasks are performed by imagining doing a specific task and classifying MI through EEG signal processing. However, it is a challenging task to classify EEG signals accurately. In this study, we propose a LSTM-based classification framework to enhance classification accuracy of four-class MI signals. To obtain time-varying data of EEG signals, a sliding window technique is used, and an overlapping-band-based FBCSP is applied to extract the subject-specific spatial features. Experimental results on BCI competition IV dataset 2a showed an average accuracy of 97% and kappa value of 0.95 in all subjects. It is demonstrated that the proposed method outperforms the existing algorithms for classifying the four-class MI EEG, and it also illustrates the robustness on the variability of inter-trial and inter-session of MI data. Furthermore, the extended experimental results for channel selection showed the best performance of classification accuracy when using all twenty-two channels by the proposed method, but an average kappa value of 0.93 was achieved with only seven channels.

In a Vehicular Ad-hoc Network Environment, roadside units (RSU), which are traffic data transmitters located statically on the roads, are the most essential infrastructural units for supporting seamless communications. In this paper, we propose an improved traffic flow based RSU allocation strategy that places the least number of RSUs at the most popular and important intersections where vehicles pass by the most frequently. We newly define the concept of intersection connectivity based on the traffic influence to an adjacent roadway, traffic flow probability, and the distance between intersections. We find RSU allocation candidates by clustering intersections based on the intersection connectivity using with Markov clustering algorithm, and then finally find RSU allocation locations according to the intersection priorities. Consequently, our strategy can place RSUs at intersections having high priorities and maximal traffic flow at the same time.

Secure aggregation of local learning model parameters is crucial for achieving privacy-preserving federated learning. This paper presents a novel and practical aggregation method that effectively combines the advantages of masking-based aggregation with those of homomorphic encryption-based techniques. Each node conceals its local parameters using a randomly selected mask, independently chosen, thereby eliminating the need for additional computations to generate or exchange mask values with other nodes. Instead, each node homomorphically encrypts its random mask using its own encryption key. During each federated learning round, nodes send their masked parameters and the homomorphically encrypted mask to the federated learning server. The server then aggregates these updates in an encrypted state, directly calculating the average of actual local parameters across all nodes without the necessity to decrypt the aggregated result separately. To facilitate this, we introduce a new multi-key homomorphic encryption technique tailored for secure aggregation in federated learning environments. Each node uses a different encryption key to encrypt its mask value. Importantly, the ciphertext of each mask includes a partial decryption component from the node, allowing the collective sum of encrypted masks to be automatically decrypted once all are aggregated. Consequently, the server computes the average of the actual local parameters by simply subtracting the decrypted total sum of mask values from the cumulative sum of the masked local parameters. Our approach effectively eliminates the need for interactions between nodes and the server for mask generation and sharing, while addressing the limitation of a single key homomorphic encryption. Moreover, the proposed aggregation process completes the global model update in just two interactions (in the absence of dropouts), significantly simplifying the aggregation procedure. Utilizing the CKKS (Cheon-Kim-Kim-Song) homomorphic encryption scheme, our method ensures efficient aggregation without compromising security or accuracy. We demonstrate the accuracy and efficiency of the proposed method through varied experiments on MNIST data.

Type 2 diabetes mellitus (T2DM) is a common, chronic disease that is closely associated with anthropometric indices related to obesity. However, no study published to date has simultaneously examined the associations of T2DM with anthropometrics, bone mineral density (BMD), and body composition variables. The present study aimed to evaluate the associations of T2DM with anthropometrics, BMD and body composition variables and to identify the best indicator of T2DM in Korean adults. The data used in this study were obtained from the Korea National Health and Nutrition Examination Survey conducted from 2008 to 2011. A total of 7,835 participants aged from 40 to 90 years were included in this study. A binary logistic regression analysis was performed to examine the significance of differences between the groups with and without T2DM, and the areas under the receiver operating characteristic (AUCs) curves were calculated to compare the predictive power of all variables. In men, waist-to-height ratio (WHtR) displayed the strongest association with T2DM (adjusted odds ratio (OR) = 1.838 [1.513–2.233], adjusted p<0.001), and waist circumference (WC) and WHtR were the best indicators (WC: AUC = 0.662 [0.639–0.685], WHtR: AUC = 0.680 [0.658–0.703]) of T2DM among all the variables. In women, left leg (LL) and right leg (RL) fat displayed strong negative associations with T2DM (LL fat: adjusted OR = 0.367 [0.321–0.419], adjusted p<0.001, RL fat: adjusted OR = 0.375 [0.329–0.428], adjusted p<0.001), and WC and WHtR were excellent indicators (WC: AUC = 0.730 [0.709–0.750], WHtR: AUC = 0.747 [0.728–0.766]) of T2DM among all the variables. In particular, the WHtR in men and LL and RL fat in women exhibited the strongest associations with T2DM, and the predictive power of the WC and WHtR was stronger than BMD, fat, and muscle mass variables in both men and women. Additionally, the predictive power of the WC and WHtR was stronger in women than in men. Of the anthropometric indices, BMD, and body fat and muscle variables, the best indicators of T2DM were WC and WHtR in both Korean men and women. The results of the present investigation will provide basic information for clinical studies of patients with T2DM and evidence for the prevention and management of T2DM.

Automatically recognizing dangerous situations for a vehicle and quickly sharing this information with nearby vehicles is the most essential technology for road safety. In this paper, we propose a real-time deceleration pattern-based traffic risk detection system using smart mobile devices. Our system detects a dangerous situation through machine learning on the deceleration patterns of a driver by considering the vehicle’s headway distance. In order to estimate the vehicle’s headway distance, we introduce a practical vehicle detection method that exploits the shadows on the road and the taillights of the vehicle. For deceleration pattern analysis, the proposed system leverages three machine learning models: neural network, random forest, and clustering. Based on these learning models, we propose two types of decision models to make the final decisions on dangerous situations, and suggest three types of improvements to continuously enhance the traffic risk detection model. Finally, we analyze the accuracy of the proposed model based on actual driving data collected by driving on Seoul city roadways and the Gyeongbu expressway. We also propose an optimal solution for traffic risk detection by analyzing the performance between the proposed decision models and the improvement techniques.

Roadside Unit (RSU) is an essential unit in a vehicular ad-hoc network (VANET) for collecting and analyzing traffic data given from smart vehicles. Furthermore, RSUs can take part in controlling traffic flow for vehicle's secure driving by broadcasting locally analyzed data, forwarding some important messages, and communicating with other RSUs, and soon. In order to maximize the availability of RSUs in the VANET, RSUs need to be fully distributed over an entire area. Thus, RSUs can make the best use of all traffic data gathered from every intersection. In this paper, we provide intersection-priority based RSU placement methods to find the optimal number and positions of RSUs for the full distribution providing with a maximal connectivity between RSUs while minimizing RSU setup costs. We propose three optimal algorithms: greedy, dynamic and hybrid algorithms. Finally, we provide simulated analyses of our algorithms using real urban roadmaps of JungGu/Jongrogu, YongsanGu, and GangnamGu in Seoul, each of which has a characteristic road style different than each other. We analyze how our algorithms work in such different types of roadways with real traffic data, and find the optimal number and positions of RSUs in these areas.