Improved networks for cloud-car communication

Title	Improved networks for cloud-car communication
Summary	Development of new communication network for more efficient, reliable and diverse traffic flow and hence improved performance driving of cars.
Keywords	Generative Adverserial Network, realtime communication, reliability, security, CAN bus technology
TimeFrame	Spring 2020
References	Generative Adverserial Nets Etschberger: CAN Controller Area Network - Grundlagen, Protokolle, Bausteine, Anwendungen
Prerequisites	Mathematical statistics, machine learning, programming
Author
Supervisor	Cristofer Englund, Eric Järpe, Ana Magazinius
Level	Master
Status	Open

As cars are getting smarter, higher demands on the communication with cloud computers are formulated. New technology (such as CAN bus) makes radiacally improved performance possible, but for accessing necessary computation capatity cloud resources are desired and then increasing amounts of data needs to be distributed via a reliable communication line. Since most or all of the information has to be delivered in realtime the design of this communication network is a delicate task.

The state-of-the-art models to this end schould be described and their pros and cons briefly discussed. A new network should be specified and constructed, possibly by means of a GAN (Generative Adverserial Network) taking into consideration how fast different parts of the communication needs to be, how it could be transmitted securly, how reliable the transmission is etc. Ideas for improvements compared to the existing solution could be suggested, implemented and tested in comparison to the stat-of-the-art solutions. New criteria for comparing the methods could be proposed and procedures for evaluation of the methods according to these new criteria could be developed. Experiments with possible techniqes to detect attacks against the security is planned. The thesis will be a part of the CyRev research project in collaboration with partners from the industry.