In this paper, a low computational burden measurement-based under frequency load shedding (UFLS) scheme is proposed to consider time delays, measurement uncertainties, and failures or losses of communication links. Using a rolling first-degree polynomial function, an iterative data-driven based approach to estimate the frequency nadir and the associated frequency nadir time is proposed. Then, a novel approach to simultaneously estimate the power imbalance and the system inertia constant is developed. To guarantee accurate estimations, the base system is selected from a set of pre-defined base systems, according to an analytical basis. Based on the above estimated parameters, a mathematical index is investigated to calculate the threshold for the generation-load power imbalance beyond which loads should be shed. The effectiveness of the proposed methodologies is examined on both, simulated and measured data from 9 confirmed historical power system disturbances in a real national power grid.