Lightweight, low-resource, Agents are installed on computers (laptops and PCs, with phones and tablets coming soon) where your users/customers are located, and are controlled centrally. They run invisibly, so that they can even be placed on real user/customer devices. The Agents don’t monitor or record what users do in any way – in fact, they completely ignore the user. Think of an Agent as a measurement instrument hosted by computer devices in offices and homes.
Agents make general performance measurements of networks and servers that host applications. They automatically transmit measurement data back to our Analytics Engine, which transforms general performance data into application-specific human experience using PQ algorithms (the research). Immediately, you will be able to quantify human experience and diagnose sources of impairment when they misbehave.
The Perception Agent exercises the network and server infrastructure by coordinating the use of standard command-line tools such as ping, traceroute and curl. From these tools (or equivalent operations), standard metrics (e.g. round-trip time, loss, goodput, etc.) are obtained and converted using a patent pending technique into Perceptual Quality scores. The technique uses complex mappings that reflect the impact of the metrics on a user's perception of application performance. The scores act as a reliable proxy for human voting.
There are a suite of parameters that can be used to adjust the coordination of the command line tools as well as to adjust the complex mappings used to produce the Perceptual Quality scores. For example, there are parameters that change the emphasis given to delay or loss metrics, or the degree of interactivity required by the application. These parameters are connected into the detailed workings of the complex mappings, and hence enable the proxy to be tuned to the needs of specific applications or user groups.