We are trying to get nearest neighbors within a user given radius. As far as I understand ST_DWithin degrades as distances are increased. The problem with using KNN is that it keeps on searching even though the radius is crossed.
For example, let's say we are searching for 3 petrol stations in 1 km. KNN keeps searching for 3 candidates and we had to filter out those outside the radius after KNN is done. This leads to unnecessary time consumption.
Is there a way that we can do nearest neighbors search within a user given radius which combines the benefit of both. Or is there an alternative better than these 2 ?