Page 798 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

2 Use Case Description

2�1 Description

In the emerging world of intelligent transportation systems, Vehicle-to-Everything (V2X)
communication is evolving to provide safer and more efficient road networks. Some of the
promising applications of V2X for wireless control use cases are Platooning, Cooperative
Transport of Goods, Cooperation between Automated Guided Vehicles (AGVs) and Stationary
Robots, etc. [1]. These applications require frequent exchange of data between vehicle sensors
and wireless networks to fulfill Quality of Service (QoS) requirements for safe and coordinated
movement.

We consider a platooning of vehicles scenario, as shown in Fig. 1, in which several vehicles are
traveling in a group, along a highway, and connected to base stations. The vehicles within the
group trail a single leading vehicle positioned ahead of them. Data from these vehicles, along
with network information from their associated base stations, is gathered at a Software Defined
Network (SDN) controller. The data includes wireless network information such as base station
transmit power, bandwidth, wireless link quality in terms of RSSI, RSRP, RSRQ, and SNR, etc,
and metadata such as vehicle-to-vehicle (V2V) data: vehicle location information, sensor data:
camera data and environment data: weather condition, traffic jam factor, etc. This extensive
data is used for making safety decisions at the Multi-access Edge Computing (MEC) to minimize
the risk of collisions or congestion.

Problem:

The current platooning solutions often rely on expensive and complex decision-making
algorithms for coordinated movement. Hence, QoS estimation in the form of reliable wireless
transmission is important for wireless control use cases. QoS estimation in non-stationary
radio environments can help to avoid any unexpected service interruptions by allowing the
network to take proactive actions. It helps vehicles to take actions such as task offloading of
on-device computation and decision making. Examples of task offloading include gaming,
object detection, and infotainment services. Proactive actions are taken based on inferences,

762