Spatio Temporal Resource Search
Addresses the problem of finding a parking slot without incurring the cost of the installing and maintaining the sensors. It deals with identifying various algorithms or approaches based on the level of information about the availability of parking slots.
Implementation Details:
The project code can be found at the below link:
What does it do?
The project aims at analyzing various algorithms on given Spatio-Temporal dataset. The dataset contains information about parking and de-parking activities for a range of a month. The given data holds about 1.5 million records. We devised and implement algorithms and examined their behavior. We then simulated an incoming car and using the given dataset, run the algorithm to determine the travel time suggested by each algorithm before getting the parking slot. The algorithms are categorized based on the level of information the algorithm has. The Three algorithms proposed are:
ALGORITHM LEVEL OF INFORMATION
Baseline No Information
Probabilistic Historical Information
Deterministic Real time Information
Baseline Approach
This approach simulates the parking slot search which is done by a driver who does not have any information of the available parking slot. The algorithm will have knowledge of the road network to search for a parking slot, and the availability of a slot in a particular block at a particular time.
Probabilistic Approach:
This approach has an historical information of parking and de-parking activities for each block and for different time frames. Using this information a predictive model is generated. This model will take a block id and time stamp as input and will return the expected number of slots for various block. The algorithm then, uses the pre-computed data finds the availability of a parking slot.
Deterministic Approach:
For this approach, we consider the real time information of parking and de-parking activities. The given dataset provides information about number of parking slots available for each block and for different time frames. The algorithm takes current block id of user and timestamp as an input. Based upon the least distance it finds a parking slot
Let's see an example:
Step 1
Consider a road network as shown, let’s say the car starts from the location shown
Step 2
The car takes a straight path based on the highest value of ratio
Step 3
The car reaches the other end of the selected block and now, again calculates ratio of the corresponding adjacent blocks
Step 4.
The car reaches the other end of the selected block and now, again calculates ratio of the corresponding adjacent blocks