Use of wearable and smart devices for real-time telemedicine and ambulatory care is gaining importance which needs reliable, accurate and robust biomedical signal analysis by specifically exploring their signatures characteristic of their corresponding physiological events. However, such analysis faces challenges due to the complex nature of these biomedical signals which often are recorded as short time series data, presence of various noises that challenges existing complexity classification algorithms such as the commonly used Shannon Entropy (SE) and other techniques which are limited by their requirement for long time series data. The purpose of this study was to demonstrate the robustness of multi scale frequency (MSF) approach for short time series that can be non-linear and/or non-stationary for various applications. Efficacy and robustness of MSF approach was tested with respect to various noises such as white, pink and brown noise using simulated sine wave and ECG data. MSF robustly estimated complexity compared to SE with various noises thereby demonstrating its efficacy for analyzing short time series signals. The results indicate huge promise for wide range application of MSF technique on a variety of time series data including biomedical signals that can be used for advanced prognosis and diagnosis of variety of diseases for wearable and smart devices suitable for telemedicine and ambulatory care.