Godkända
Reduktion av bandbredd genom Edge Computing i uppkopplade bilar
Mikael Jarfors (2013) och Axel Rosén ()
Start
2018-06-25
Presentation
2018-11-08 10:15
Plats:
E:2349
Avslutat:
2018-11-24
Examensrapport:
Sammanfattning
Edge Computing is a paradigm for connected devices where processing is done in the nodes themselves, instead of in a central server unit. An example of this is in a surveillance camera scenario, where you may have a certain total bandwidth of, for example 10 units. If each connected camera streams costs 1 unit, then you can have 10 cameras in total. However, it may be unnecessary for the cameras to stream constantly. Therefore, if a simple calculation or interpretation is done inside the camera hardware, such as checking if there is movement in the frame, and only then allow the camera to stream. This methodology would allow for more cameras than 10 to be able to use the network, since each camera would not need its full bandwidth in each case, this is called Edge Computing. Volvo has a data collection service in their cars, meaning that the car itself has numerous sensors that are constantly being collected into local storage. These sensors output data in the form of data streams, for example the engine/power-train would have the current amount of output power streamed to this data collection service. The data collection service also functions as a server for processes which want to use the information in the sensors, such as if you would want to observe the current power output from the engine, then the service could subscribe to the power-train sensor output. However, having many services subscribing to the data collection service will cause stress on the data collector itself. Therefore, new solutions on how to manage the subscriptions are managed need to be created, such as grouping the same subscriptions into only pipeline of information, or simple interpretation of the data to allow of less bandwidth intensive connections. The main goal of the masters thesis is to investigate whether it is possible to use Edge Computing in order to optimize the way Volvo manages the way the data collection services work in their connected cars and how to optimize it for their use case. The main edge computing use case is to optimize on generic processing of diagnostic data eciently in the car.
Handledare: Lars Larsson (EIT) och Maria Kihl (EIT)
Examinator: Christian Nyberg (EIT)