Sliding-Window Approach for Improving Response Time of Mission-Critical Applications

Abstract

Cloud computing is reaching its limits when it comes to mission-critical applications. Edge computing, on the other hand, attempts to place computing capabilities closer to the user equipment device, minimizing response time and increasing reliability. In order to take full advantage of the edge and to offer a broad service coverage, a wide range of mission-critical applications use mobile edge computing. However, one of the challenges that affects reliability in mobile edge computing is the border problem. The border problem typically occurs when the mobile user equipment crosses the border between two points of access. During this condition, the response time increases and becomes less predictable, which severely hurts the reliability of mission-critical applications. This work tackles this problem by developing a sliding-window approach, which is to maintain multiple connections between a user equipment and several edge node servers. The sliding-window reduces the adverse effects produced by the border problem in mobile edge computing, ensuring a more predictable response time. The proposed approach is implemented in the Street-Edge Hazard Notification System (SEHNS), a mission-critical application in the domain of automotive safety that disseminates hazards information among intelligent vehicles in real-time. Initially, an analysis of the border problem cases is carried out. Then, the performance of SEHNS is examined in each of the cases, with and without sliding-window. Finally, the approach is implemented in a realistic scenario where two vehicles follow each other along a road with several hazards. Results show that the sliding-window approach reduces the worst case response time by about 50%.