An Experimental Framework for the Analysis and Validation of Software Clocks

The experimental evaluation of software clocks requires the availability of a high quality clock to be used as reference time and a particular care for being able to immediately compare the value provided by the software clock with the reference time. This paper focuses i) on the definition of a proper evaluation process and consequent methodology, and ii) on the assessment of both the measuring system and of the results. These aspects of experimental evaluation activities are mandatory in order to obtain valid results and reproducible experiments, including comparison of possible different realizations or prototypes. As case study to demonstrate the framework we describe the experimental evaluation performed on a basic prototype of the Reliable and Self-Aware Clock (R&SAClock), a recently proposed software clock for resilient time information that provides both current time and current synchronization uncertainty (a conservative and self-adaptive estimation of the distance from an external reference time).