This paper presents an algorithm to re alize the detection and recognition of the optical fiberintrusion signals.It is decomposed and mapped on the embedded processor consisting of field-programmabl e gate array (FPGA) and digital signal processing (DSP) according to the processing time and operati on characteristics of the algorithm.The treatment process is divided into three parts:firstly,the original signals use 3Hz high-pass filtering,the high-pass filtered signal is divided into two parts respectively by two DSP in parallel to save time cost,and all subsequent operations are in a pa rallel mode.The filtered data are standardized and compared with the detection threshold.If th e result is greater than the thershold,the signal is judged as vibration siganl and marked as 1;if n ot,judged as the false alarm and marked as 0,and then the single path detection results are obtai ned.The data used for recognition is extracted based on the above detection resu lts.Finally,the data input recognition module is used for feature extraction of mechanical,walking,di gging signals, and the recognition results are obtained.The experimental results show that the algorithm can effectively detect and recognize the optical fiberintrusion signals,improve the operation speed and satisfy the real-time performance request of the optical fiber prewarning s ystem (OFPS) to detect and recongnize signal type.