The absence or a low level of testing for COVID-19 can cause a significant risk of being misled on the actual nature of outbreak, while a large-scale testing and quick testing turnaround are costly. There-fore, it is essential to understand how human Social Distancing (SD) behavior could aid in limiting the disease spread, and to find a systematic testing strategy that allows to fine-tune human behavior to slow the epidemic in the most efficient way. We propose to utilize an SIR model which takes into account both mild and symptomatic diseases, and is implemented on a real-life social network, based on the synthetic data generated by Simfrastructure. The network is endowed with demographics consistent with individuals’ activities, locations and interaction with others on a second-by-second basis. The nodes of the network are synthetic people and the edges are contacts via different activity locations (home, work, school, shopping center, etc). We update this contact network so it only includes edges causing spread of COVID-19, and then incorporate self-regulated SD as a function of the infection prevalence among social contacts and infection progression, with the aim of optimizing the use of limited testing efforts.