We investigate the k-error linear complexity of p 2 -periodic binary sequences defined from the polynomial quotients (including the well-studied Fermat quotients), which is defined bywhere p is an odd prime and 1 ≤ w < p. Indeed, first for all integers k, we determine exact values of the k-error linear complexity over the finite field F 2 for these binary sequences under the assumption of 2 being a primitive root modulo p 2 , and then we determine their k-error linear complexity over the finite field F p for either 0 ≤ k < p when w = 1 or 0 ≤ k < p − 1 when 2 ≤ w < p. Theoretical results obtained indicate that such sequences possess 'good' error linear complexity.

We investigate the k-error linear complexity of pseudorandom binary sequences of period p r derived from the Euler quotients modulo p r−1 , a power of an odd prime p for r ≥ 2. When r = 2, this is just the case of polynomial quotients (including Fermat quotients) modulo p, which has been studied in an earlier work of Chen, Niu and Wu. In this work, we establish a recursive relation on the k-error linear complexity of the sequences for the case of r ≥ 3. We also state the exact values of the k-error linear complexity for the case of r = 3. From the results, we can find that the k-error linear complexity of the sequences (of period p r ) does not decrease dramatically for k < p r−2 (p − 1) 2 /2.

Encrypted traffic classification plays a vital role in cybersecurity as network traffic encryption becomes prevalent. First, we briefly introduce three traffic encryption mechanisms: IPsec, SSL/TLS, and SRTP. After evaluating the performances of support vector machine, random forest, naïve Bayes, and logistic regression for traffic classification, we propose the combined approach of entropy estimation and artificial neural networks. First, network traffic is classified as encrypted or plaintext with entropy estimation. Encrypted traffic is then further classified using neural networks. We propose using traffic packet’s sizes, packet's inter‐arrival time, and direction as the neural network's input. Our combined approach was evaluated with the dataset obtained from the Canadian Institute for Cybersecurity. Results show an improved precision (from 1 to 7 percentage points), and some application classification metrics improved nearly by 30 percentage points.

No abstract