Objective: This papers proposes an initial proof-of-concept validation for a method that estimates} the respiratory effort during pressure support ventilation, based solely on \WPL{the signals provided by a standard commercial ventilator. The study is carried out on a test bench with an Active Servo Lung 5000 mechanical simulator. Methods: The asymptotic behavior of the respiratory mechanics during expiration is exploited to analyze its identifiability (the fact that two sets of respiratory mechanics parameters necessarily generate distinct signals). Furthermore, we propose a method of estimation of the patient's respiratory muscle pressure, based on the degree of smoothness of a signal constructed from the airway pressure and flow measurements. Results: The proposed method was validated on a commercial ventilator, under a wide range of respiratory mechanics parameters and respiratory efforts. Conclusion: Our approach provides a non-invasive estimation of the respiratory effort, which allows for accurately detecting over- and under-assistance. Significance: Compared to existing muscle pressure estimation methods, our approach clarifies the conditions under which the respiratory effort of a patient can be estimated and offers a simple way to compute it in real time, without using invasive devices or occlusion-based techniques.