Parking algorithm designed to influence the demand for on-street parking

Nir Fulman, Itzhak Benenson

Department of Geography and Human Environment

Porter School of Environment and Earth Science, Tel-Aviv University

Searching for curb parking is an all-or-nothing game. Just recall from your own experience – parking is either freely available near your destination, or the curb is full and you slowly lose hope circling around for lengthy amounts of time, and later blame yourself for the lost time when you finally park at a distant parking lot. Why didn’t you drive to this lot at once?... In your heart you know why – there were a few times you’ve gotten lucky, somebody had just left and the spot was yours. And this could have happened again!

Our pursuit of parking 'luck' is quite detrimental to urban traffic. Frequently cited hearsay is that searching for parking can attribute to 30% of the CBD traffic in peak morning arrival time. We failed to track the source, but a twice lower estimate of 15% easily covers the entire gap between free traffic flow and heavy congestion. How can we prevent frustrating circling and convince ourselves that, if we don’t see a free spot near the destination, the most time-preserving option is to go to the nearest parking lot?

At one point, we thought an app that directs drivers to parking lots that are not fully occupied would be valuable. While this would be useful, drivers would still only leverage the capabilities once they decided to give up hope on finding nearby street parking – which defeats the purpose of lowering the congestion and changing the behavior in the first place.

Circling for on-street parking has its incentives – as it is much cheaper for short term parking and usually more convenient than a nearby parking structure. However, too many of us compete for this publicly owned resource and the demand outweighs the supply of curb space. The idea of the “social” parking, where members of a club inform other members that they are going to leave, doesn’t work not either, as other 'circling' drivers will get to the opening spot first. The only way to encourage drivers to stop adding to congestion and skip hunting for on-street parking, is to lower the desirability and price space like any other property – in response to demand.

Demand-responsive parking prices are becoming increasingly more popular in North America in recent years. In San Francisco and Los Angeles, sensors measure the exact occupancy rate of each street link throughout the day and then increase parking prices again and again until occupation rate decreases to a level of 80-85% that is, one free spot to 4 - 6 occupied ones. The problem is that the sensor system costs millions of dollars to install and operate. Our studies (Fulman & Benenson, 2018; Fulman & Benenson, 2019) clearly indicate that parking occupation rate, at any hour of a day, can be predicted based on standard spatial information (GIS layers) of urban streets, buildings, and parking lots that are available in every municipality. If so, parking search and drivers’ reaction to prices can be simulated algorithmically and expensive sensors become redundant.

Recently published in the IEEE ITS Magazine (Fulman, Benenson, 2018), our approach to fast and frugal estimation of parking demand and supply algorithm makes possible to establish parking prices that would guarantee any desired occupation rate, 80% or higher. It has already been successfully tested in the Israeli city of Bat Yam and resulted at socially acceptable curb parking fees (Figure 1).

Figure 1. Bat Yam parking prices, by street links, that would preserve 80% occupation (Fulman, Benenson, 2018)

Figure 1. Bat Yam parking prices, by street links, that would preserve 80% occupation (Fulman, Benenson, 2018)

The algorithm is flexible and can accommodate ‘free of charge’ permits or discounts to resolve socio-economic inequalities. It can also account for local circumstances that influence demand and supply, such as a ball game or a new parking lot. Parking prices can be established at any resolution – street links, neighborhoods, urban regions. However, larger units are easier for the drivers to comprehend.

With adaptive prices, drivers who really need to park close to their destination could always find a spot available. However, to pay less, they would opt to park further away or in a lot/parking structure. A mobile app could inform drivers of prices and let them choose parking that would be always available, before arriving to the area. It would prevent people from circling for parking in city centers and nobody will be charged more than necessary to ensure the desired occupancy rate is achieved.

Fulman, Nir, and Itzhak Benenson. "Establishing Heterogeneous Parking Prices for Uniform Parking Availability for Autonomous and Human-Driven Vehicles." IEEE Intelligent Transportation Systems Magazine 11.1 (2018): 15-28, available at:

Fulman, Nir, and Itzhak Benenson. "Approximation of Search Times for On-street Parking Based on Supply and Demand." arXiv preprint arXiv:1806.10874 (2019), available at:

Commenting is disabled on this post.