Can Spark structured streaming run a window operation if a DataFrame does not have a timestamp column? If yes, then how?

Spark window operations are not supported on non-timestamp-based columns and, so, it is not possible to run these operations on such columns.

Also, if the source is Kafka and you don't have a timestamp column, then it can get an event timestamp when data arrives at the system, and that timestamp can be used for window operations, for instance, all data arriving within a window frame of 10 mins.