An approach is provided for sampling crowd sourced data. The approach selects an sampling node from a set of crowd nodes. The sampling node receives a data acquisition request from a data collector and receives data from the set of crowd nodes with the data being responsive to the data acquisition request. The received data is processed by the sampling node to reduce redundant data as defined by the data acquisition request. An acquired data message block is generated and transmitted from the sampling node to the data collector.