The Global Navigation Satellite System (GNSS) serves people as an essential precise time synchronization source. However, to provide public services on a global scale, GNSS civil signals need to have open structures. And their power is much lower than thermal noises at the receiver end. These make spoofing a major threat and GNSS timing users are easy to be deceived by a kind of spoofing known as the Time Synchronization Attack (TSA). The real-world damage of TSA is proved by researchers and demands of TSA defense methods emerge. Typical anti-spoofing methods can bring some degrees of security, but they mainly focus on physical parameters that do not affect timing solutions. So attackers always have possibilities to cheat these methods. A more robust way is to directly monitor key timing results, like clock biases and clock drifts. Researchers have come up with several methods of this kind, but their requirements of complex computations and incessant mitigations may bring difficulties to actual usages. To complement these deficiencies, this paper proposes a TSA detection and discrimination method based on the abnormal similarity between signals from different satellites. Specifically, we calculate clock drifts from each satellite signal separately and monitor the correlation values of different clock drifts time-differenced sequences. As TSA signals need to avoid sudden changes in timing results and keep actual position solutions, spoofing is ought to maintain a same changing rate of pseudorange differences among all spoofing-authentic signal pairs. This additional similarity will greatly enlarge the correlation value, thus, TSA is revealed. After detected and discriminated by the proposed method, spoofing affects will be much easier to erase. We achieve this method by monitoring timing results only and requiring few complex computations, so it will be a sufficient complement to off-the-shelf real-world TSA defenses. Our motivations and method effects are verified on an open test-bed.