Abstract

Tactile Internet is a next generation Internet that allows the transmission of haptic sensations in addition to audio and video content.
It is expected to enable new latency-sensitive and critical use cases, such as remote phobia treatment, tele-surgery and autonomous driving. However, the current networking infrastructure cannot ensure the strict requirements that come with Tactile Internet, namely ultra-responsiveness and ultra-reliability.
\indent Edge computing can help in solving this issue. While Cloud Computing offers powerful computing resources at distant data centers, Edge computing provides resources closer to the end user. To this end, computations can be offloaded from the cloud to the edge to obtain lower latency. In addition, as the Edge itself may prove to be insufficiently close to the end users’ devices in some cases, it can be augmented with Mobile Ad-Hoc Clouds. The Mobile Ad-hoc Clouds refer to a group of mobile devices located at the immediate vicinity of the end users, offering their available resources for computation, leading therefore to a reduced latency. Nevertheless, the design and implementation of an architecture based on edges augmented with mobile ad-hoc clouds for Tactile Internet raises several challenges. Firstly, a tactile internet-based architecture for remote phobia treatment should allow the exchange of auditory, visual and haptic information to ensure the efficiency of the therapy. Secondly, the end to end latency should be in the order of a few milliseconds to avoid “cyber-sickness”.
\indent This thesis provides a case study of edge augmented with mobile ad-hoc clouds architecture for remote phobia treatment. The contributions are threefold. First, a software architecture for remote phobia treatment is designed for an edge augmented with mobile ad hoc clouds environment. Second, a proof of concept prototype for the proposed architecture is implemented and evaluated using a set of haptic devices, which include the HTC Vive VR headset, the Leap Motion hand tracking device, as well as the Gloveone haptic glove. Third, a set of experiments consisting of placing the components in the different layers (i.e. Cloud, Edge and Mobile Ad-hoc Cloud) were conducted, which allowed an evaluation of the impact on performance. A set of high-level interfaces were introduced to allow communication with the heterogeneous devices. The design of the architecture as a set of software modules allows the reusability of the architecture.