This project attempts to demystify the nuanced complexities as observed from a prototype software model that can be used to initiate the learning component of a neural network architecture assumed to be in control of content from public server. In this case any remote CARLA server.
Vehicles are simulated using AUTOCAST system which produces the intersection scenario. Vehicle drivers are implemented as artificial agents tasked with driving carla car models. If vehicle is not CH and granted access to cluster then it will replicate its memory to all the nodes on the cluster Otherwise, the vehicle will have to retry again. The state of the vehicle will also change
CARLA lacks network modules to enable simulation of communication. In this research, NS-3 is proposed to be used as the network simulator. Though this work hasn't initially started as of date (Nov 15, 2020). Instead of using NS-3 though, MQTT will be used as the transport protocol between the
For this current version of the application, Emna Daknou fast clustering algorithm will be implemented.
Emna Daknou paper, the system has the following stages:
- Discovery
- Cluster Formation
- Cluster Head Election
- Recovery
What is hard is estimating the states of a node.
"""