Figure 4. Flow of designed estimation technique
A. Steps in Building the Model
In this paper. our efforts on building a model, and validating its efficacy in the context of estimating distance(ranging) and intensity of an explosion event comprises of four steps as illustrated in Figure 4
Step 1: Event Separation: This is a pre-processing step of training. Recall that if a smartphone continuously senses ambient vibrations and stores all readings. then the storage overhead will be significant. As such, in real-life scenarios, we need a technique that will allow the phone to retain only those vibration readings that correspond to an explosion while filtering out the rest. We implement the technique to do so as an algorithm that runs in real-time on the phone as an app. The core rationale of our technique is to compute in run-time, the ratio of averages of vibration readings within a short-term sliding window (denoted as STA) to the long term sliding window (denoted as LTA).