An industrial plant has hundreds or even thousands of control loops. The performance of control loops is important to ensure high product quality and low cost of the product in such plants. In general, oscillatory variables are one of the main causes for poor performance of control loops. Therefore, detection and diagnosis of the causes of oscillations in control loops is very vital for maintaining the product within the desired quality limits. Identification of such source signatures is a very important task in assessing closed loop control performance. In the current study, Non-negative matrix factorization (NMF), a multivariate source identification technique has been utilized for oscillation detection and isolation in closed loops. Further, the issues present in the existing source identification techniques for oscillation diagnosis is addressed. After identifying the control loop which is the cause for plant-wide oscillation, it is necessary to identify the source of oscillation in the particular control loop. In general, poor controller tuning, control valve stiction, poor process and control system design, and oscillatory disturbances are sources of oscillations in a closed-loop control system. A frequency domain algorithm based on Hilbert-Huang transform is developed to identify the oscillations produced by the sticky valve and quantify the amount of stiction present in the control valve apart from pointing out the drawbacks in the existing bicoherence analysis method for detection of valve stiction. Besides valve stiction, delay mismatch between the model and plant is a major source of oscillation. In the current work, a novel approach has been proposed to identify the delay present in the closed loop single-input single-output systems (SISO) using information theory.