Time Series Rolling Window Mean with Nearest Points - sql

Question

I want to implement a function in that could compute rolling mean and median on a time series data using nearest neighbour points and I am wondering how I can do it in SQL.
To illustrate what I am trying to do:
Assuming I have
data containing information on parked vehicles across all carparks in USA everyday (From 2022-01-01 till 2022-07-18) as well as their exact parking location (latitude longitude) and the cost of the vehicle.
Vehicle Tracking System containing 10 unique Tracker ID, Date Installed, its latitude and longitude in a particular carpark
Example of the first data:
Date, vehicle ID, cost of vehicle (in $), latitude, longitude
2022-07-18, 1, $4000, 3.15, 100.15
2022-07-18, 2, $3000, 3.11, 100.13
2022-07-17, 3, $5000, 3.12, 100.10
...
Example of second data:
Tracker ID, Date Installed, latitude, longitude
A, 2022-07-18, 3.14, 100.12
B, 2022-07-20, 5.10, 105.14
For each Tracker, I want to know what is the mean and median cost of the nearest 5 vehicles parked within 1km of its vicinity based on the past 5 days parking data, as well as past 10 days parking data from its date of Installation (Vehicles can freely move in and out of a carpark anytime)
Example of how my output should be:
Tracker ID, Mean_Cost_5, Median_Cost_5, Mean_Cost_10, Median_Cost_10
Tracker 1, $4000, $4000, $7000, $5000 (Using carpark data from 5 and 10 days before 2022-07-18)
This should be done for each of the 10 Trackers.
How can I implement it in SQL?