The solutions to the problem of the tracking a wireless node is approached conventionally by: 1) proximity detection; 2) triangulation; 3) scene analysis methods. In these, scene analysis method is simple, accurate and less expensive. Indoor localisation technologies need to address the existing inaccuracy and inadequacy of global positioning-based systems (GPS) in indoor environments (such as urban canyons, inside large buildings, etc.). This paper presents a novel indoor Wi-Fi tracking system with minimal error in the presence of barrier using Bayesian inference method. The system integrates an android app and python scripts (that run on server) to identify the position of the mobile node within an indoor environment. The received signal strength indicator (RSSI) method is used for tracking. Experimental results presented to illustrate the performance of the system comparing with other methods. From the tracked nodes, a theoretical solution is proposed for finding shortest path using Steiner nodes. © 2020 Inderscience Enterprises Ltd.