a new method to detect and mitigate time synchronization attack on gps clock offset based on linear regression predictor

Since some gps signals are unencrypted and available to the public, they are subject to some interference that can cause errors or deception in gps signals. we discuss the effect of time synchronization attack (tsa) on the clock offset of commercial gps receivers with crystal and atomic oscillators. then, a new innovative algorithm based on linear regression prediction (lrp) is proposed. with this algorithm, we can predict the clock offset and compare it with the clock offset estimated by the gps receiver. in this case, if a spoofing attack occurs, we can detect it and then correct the clock offset. the efficiency of the proposed algorithm has been compared with other methods such as predictive multi-layer perceptron neural network (mlp nn). a proper and deep understanding of the clock offset behavior is the key to providing a simpler algorithm with greater efficiency and accuracy than earlier methods. due to its simplicity and the need for less training data, the proposed algorithm has a 3-4 times higher learning speed than the other predictive methods. less computational complexity, more prediction horizon, and high prediction accuracy are other features of the proposed lpr algorithm.

a new method to detect and mitigate time synchronization attack on gps clock offset based on linear regression predictor

since some gps signals are unencrypted and available to the public, they are subject to some interference that can cause errors or deception in gps signals. we discuss the effect of time synchronization attack (tsa) on the clock offset of commercial gps receivers with crystal and atomic oscillators. then, a new innovative algorithm based on linear regression prediction (lrp) is proposed. with this algorithm, we can predict the clock offset and compare it with the clock offset estimated by the gps receiver. in this case, if a spoofing attack occurs, we can detect it and then correct the clock offset. the efficiency of the proposed algorithm has been compared with other methods such as predictive multi-layer perceptron neural network (mlp nn). a proper and deep understanding of the clock offset behavior is the key to providing a simpler algorithm with greater efficiency and accuracy than earlier methods. due to its simplicity and the need for less training data, the proposed algorithm has a 3-4 times higher learning speed than the other predictive methods. less computational complexity, more prediction horizon, and high prediction accuracy are other features of the proposed lpr algorithm.