Algorithms

We consider several scheduling algorithms, which will be presented in the following paragraphs. A complete description of the algorithms can be found in the project's paper.

• The first algorithm is the Static Scheduler. This is the default scheduling algorithm of SUMO. This algorithm gives each lane in each junction a constant amount of green-light time, in a round robin way. There are two main observations to this scheduler: it schedules each junction independently of the other junctions. In addition, it does not use the input from the detectors on the lanes. Therefore, this scheduling algorithm corresponds with the scheduling algorithm of a non-smart city (regular city).
• The second algorithm, is the Max Occupancy Scheduler. This is a relatively simple algorithm. At each step it chooses the most occupied lane to be scheduled next, i.e. grants a green light to the lane with the most cars. There are three important observations to this algorithm. First, this algorithm schedules each junction independently of the other junctions. Second, this algorithm does use the information from the detectors in order to decide which lane is the most occupied. Thus this is the simplest algorithm that suits the definition of a smart city. Third, this algorithm may cause starvation of a specific lane. If a certain lane is never the most occupied lane in a junction it will never be given a green light.
• Finally, the third algorithm is called the Advanced Scheduler. It is the best one we have, and also the most sophisticated one. The idea is to maximize the throughput of the traffic under the junction and in addition, to prevent starvation to lanes with less traffic. In order to achieve that the traffic lights are scheduled in a round-robin way. Each junction is scheduled independently of the other junctions.