Cooperative Caching in Vehicular Networks - Distributed Cache Invalidation Using Information Freshness

Recent advances in vehicular communications has led to significant opportunities to deploy variety of applications and services improving road safety and traffic efficiency to road users. In regard to traffic management services in distributed vehicular networks, this thesis work evaluates managing storage at vehicles efficiently as cache for moderate cellular transmission costs while still achieving correct routing decision.

Road status information was disseminated to oncoming traffic in the form of cellular notifications using a reporting mechanism. High transmission costs due to redundant notifications published by all vehicles following a basic reporting mechanism: Default-approach was overcome by implementing caching at every vehicle.

A cooperative based reporting mechanism utilizing cache: Cooperative-approach, was proposed to notify road status while avoiding redundant notifications. In order to account those significantly relevant vehicles for decision-making process which did not actually publish, correspondingly virtual cache entries were implemented. To incorporate the real-world scenario of varying vehicular rate observed on any road, virtual cache entries based on varying vehicular rate was modeled as Adaptive Cache Management mechanism.

The combinations of proposed mechanisms were evaluated for cellular transmission costs and accuracy achieved for making correct routing decision. Simulation case studies comprising varying vehicular densities and different false detection rates were conducted to demonstrate the performance of these mechanisms. Additionally, the proposed mechanisms were evaluated in different decision-making algorithms for both information freshness in changing road conditions and for robustness despite false detections.

The simulation results demonstrated that the combination of proposed mechanisms was capable of achieving realistic information accuracy enough to make correct routing decision despite false readings while keeping network costs significantly low. Furthermore, using QoI-based decision algorithm in high density vehicular networks, fast adaptability to frequently changing road conditions as well as quick recovery from false notifications by invalidating them with correct notifications were indicated.