Signal analysis for a new alarm panel

Problem

A security company developed a new generation of alarm panels, more modern, faster and with new features. Two months into the project it started installing them with a few selected customers so they could test them and report incidents and improvements for future versions. The signals each panel sends needed to be understood and analysed with data: a complex structure whose meaning changes depending on how it starts.

Approach

• Sessions with the business to understand what each part of the signal means, based on documentation of more than 100 pages.
• T-SQL procedures to extract, transform and analyse each signal: turning a string of more than 50 characters into readable data (alarm status, user who sent it, previous status, time, local or remote origin).
• Rebuilding each panel’s state: depending on the signal, the previous and following signals had to be checked to confirm each state change, with more than 15 possible states and transitions that depend on the previous state.
• A team of 2 building a tree of possibilities, with dozens of meetings with the business to check the cases the initial documentation did not cover.
• A two-month project, with changing scope and a lot of trial and error until each case was closed.
• More than 40 real use cases of state changes, documented for the business.

Result (qualitative)

The business and the panel engineers got, for the first time, a complete map of how an alarm changes state. With it they identified transitions with unnecessary intermediate steps, which added a few seconds to simple tasks and, on some models, could leave the panel in a mode the user had not asked for. From there they implemented improvements to ensure correct operation and raise quality and performance.

Skills demonstrated

Advanced T-SQL with stored procedures, analysis and interpretation of complex data, requirements gathering with different stakeholders, joint validation with the business and the engineering team, teamwork.