A non-stationary signal is characterized by statistical properties that change over time, making it challenging to analyze using traditional methods that assume stationarity. This type of signal often requires advanced techniques like time-frequency analysis to accurately capture its dynamic behavior.