Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Anonymous communication networks allow users to send information anonymously over the Internet. These systems often consist of an overlay network of routers. A message that enters the system is passed through a number of these routers and relay them further until the message is delivered to the recipient. To ensure anonymity, the system aims to hide the link between sender and recipient of a message by preventing an attacker to identify a user among a set of other users. This set of users is commonly referred to as the anonymity set (Danezis et al. 2003). The size of the anonymity set can be used as a measure for anonymity (Serjantov et al. 2003). One such system is based on a server called a mix, which focuses on indistinguishably transforming batches of messages. Routing messages through a network of mixes is an effective method to blur the destination of each message from any unwanted observer. However, these networks heavily depend on methods that are able to combine efficiency, anonymity, and resilience to pass messages through multiple relays. These routing strategies generally have to deal with a trade-off between performance and security. A mix cascade (Chaum 1981) is effective in keeping large anonymity sets, but it is not scalable and prone to router failure. Free-route mixes on the other hand are more resilient but are more resource demanding (Shirazi et al. 2016). A lack of performance is directly related to reduced anonymity of a system, because unscalable systems force one to use small anonymity sets to get reasonable results. This project will show how the use of a protocol for multi-party routing using a stratified topology (Shirazi et al. 2017) can increase the scalability and robustness of mid latency mix networks. Not only does multi-party routing offer a large degree of control over the size of the anonymity set, it also provides solid resilience to router failures. The protocol thereby presents a conciliatory routing strategy for anonymous communication.

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Monitoring the jump height of volleyball players during training or a match can be important in preventing injuries, measuring performance or determining the level of fatigue. On the one hand, current methods require considerable workload and cost, such as video analysis. On the other hand, companies like VERT (VERT, Mayfonk Athletic, FL, USA) have already developed a device that can measure various parameters of jumps, including jump height. They, however, do not provide any insights into the algorithm they use and fail to perform well in matches or training. This calls for another method to determine the jump height of volleyball jumps. This study investigates the prediction of the jump height through machine learning techniques based on 3-axial accelerometer and gyroscope data from an Inertial Measurement Unit (IMU) worn by volleyball players. Using this data from volleyball players performing various jumps, different neural network architectures including Feedforward Neural networks (FF NN), Long Short-Term Memory (LSTM) networks, and Convolutional Neural Networks (CNN) were analyzed. The data consisted of 27 subjects that performed different types of jumps during three volleyball training sessions. Results demonstrate the superior performance of deep learning models, particularly FF NN, surpassing the conventional extreme gradient boost (XGBoost) regressor, with an R-squared score of 0.877 in predicting jump height, which is the highest overall R-squared score of the different models. In conclusion, these results show the potential of deep learning techniques to accurately predict the jump height of volleyball jumps.