Cloud-Based Pedestrian Road-Safety with Situation-Adaptive Energy-Efficient Communication
Research output: Contribution to journal › Article › Scientific › peer-review
- VTT Technical Research Centre of Finland
Pedestrian detection using wireless communication complements sensor-based pedestrian detection in driverless and conventional cars. This fusion improves road-safety particularly in obstructed visibility and bad weather conditions. This paper seeks developing such wireless-based vehicle-to-pedestrian (V2P) collision avoidance using energy-efficient methods and non-dedicated existing technologies namely smartphones (widespread among pedestrians and drivers), cellular network and cloud. Our road-safety mobile app can be set to driver mode or pedestrian mode. This app frequently sends vehicle and pedestrian geolocation data (beacons) to cloud servers. Cloud performs threat analysis and sends alerts to road users who are in risky situation. However, constant pedestrian-to-cloud (P2C) beaconing can quickly drain smartphone battery and make the system impractical. We employ adaptive multi-mode (AMM) approach built on situation-adaptive beaconing. AMM reduces power consumption using beacon rate control while it keeps the data freshness required for timely vehicle-to-pedestrian collision prediction. AMM runs on cloud servers and commands the mobile apps to change P2C beaconing frequency according to collision risk level from the surrounding vehicular traffic. City-scale mobility simulation demonstrates energy efficiency of our approach. We evaluate battery lifetime according to geolocational variations over the city map. Results show that road-safety system imposes a small mean overhead on smartphone battery's state-of-charge. Furthermore, our evaluation of computation and network load shows feasibility of running such road-safety systems on conventional cellular networks and cloud providers. We use server-side prototype experiment to estimate minimum cloud resources and cloud service costs needed to handle computation of city-scale geolocation data.
|Number of pages||18|
|Journal||IEEE Transactions on Intelligent Transportation Systems|
|Publication status||Published - 21 Jul 2016|
|MoE publication type||A1 Journal article-refereed|