In communication systems, embedded devices rely on encryption techniques to ensure the security of data transmission. The leakage of side-channel information during the encryption process provides potential analysis objects for attackers, which may lead to the exposure of encryption keys. However, in practical scenarios, due to the limited control that attackers have over the target devices, they often face a small sample problem, which poses challenges for profiling-based side-channel analysis (SCA). To address this issue, this paper proposes a novel cross-leakage profiling side-channel analysis method, denoted as CLeP, aimed at revealing the potential leakage threats that communication devices still face with limited data samples. We evaluate this method using the AES-128 algorithm, considering that a captured trace consists of 10 rounds of encryption, with each round containing 4 operations, and each operation involving 16 bytes. Notably, the leakage from the round key addition and byte substitution operations is the most significant. Therefore, we decompose a single trace into 336 operation values to derive the corresponding leakage features for each operation. Additionally, we introduce an autoencoder (AE) and a maximum mean discrepancy (MMD) algorithm to constrain the differences between the two types of operations. Our experimental results demonstrate that the CLeP method significantly enhances attack performance in small sample scenarios.