Wireless Sensor Networks (WSNs) can be viewed as energy constrained database systems, and join query processing is a very important topic in the field of sensor based systems. Join query for WSNs in a multi-dimensional and continuous way has not been widely explored, as well as is not energy efficient enough. In this paper, we proposed a continuous Multi-attribute Join Query Processing (MJQP) within latest sampling periods for WSNs based applications. We developed a filter-based scheme to discard non-joining tuples, which the center points of filters are identified and updated. Besides, we design an optimized solution to reduce the transmission of non-joining tuples, which is very benefit on energy efficiency. Experiments on real-world data set show that our methods outperform the centralized algorithm.