A vehicle area network (VAN) is a local area network deployed onboard a moving vehicle, e.g., a train, bus, or a private car, to provide high-speed Internet access to the passengers. A VAN may face temporary network disconnection when the vehicle passes through challenging radio environments, e.g., deep tunnels. Such disconnections disrupt the ongoing network services causing passenger dissatisfaction. Given that tunnel locations are known, we propose to use geographical knowledge in the

more »

... on control function of the VAN to reduce the probability of service disruption. A geographic admission controller (GAC) rejects a new call request based on the vehicle's current location if it can be determined that the vehicle would enter a tunnel within a short period of time. By rejecting a new call prior to entering a tunnel, GAC reduces the probability of an admitted call to be disrupted, but it does so at the expense of increasing probability that a new request would be blocked. For a quantitative investigation of the trade-off between these two important probabilities, we propose a 2-D Markov Chain model. Using the model, we derive both probabilities as a function of the time interval from the moment the GAC starts rejecting new requests until the vehicle actually enters the tunnel. This time interval is a configurable parameter which controls both probabilities. The model is validated using simulation. Our model reveals that the disruption probability can be reduced quadratically with a linear increase in the blocking probability.