Locational marginal prices (LMP) are the dual variables associated with the power balance constraint of the optimal power flow (OPF) problem. Prior works on LMP calculation and forecasting shows that LMP is composed of generation cost, marginal loss cost and distribution network (DN) congestion cost. The focus of this work is to estimate the congestion component of LMP for radial DNs with only local parameter measurements. The true congestion component of LMP is derived by solving an OPF without penalizing the DN loss or generation cost. The estimated component of the LMP, referred to as the flexibility activation signal (FAS), is derived as a function of the nodal voltage profile at the point of common coupling (PCC) and the line loading of incoming and outgoing branches from PCC. The estimated and true congestion components of LMP for all nodes are analyzed via time series analysis using Granger causality and sensitivity and specificity of estimation output. Numerical case studies provide statistical evidence that voltage times series can be utilized to estimate congestion component of DN OPF. Further, we also show that with local parameter measurements such as line loading and/or voltage measurements can accurately estimate DN congestion incidents with an accuracy exceeding 95%. Although, localized congestion estimation is not entirely accurate, however, such a method is privacy-preserving, desired for autonomous operation without the need for centralized communication. Thus, proposed decentralized FAS estimation is inherently cyber-secure and resilient to external attacks, while mitigating DN congestion.