Recent increases in homelessness in the United States have been described as a nationwide emergency. The negative impacts of homelessness on communities and individuals are well-established, including significant impacts to the environment, health and safety (e.g. fires, sanitation crisis). To address the effects of increasing homeless populations, particularly in cities on the west coast of the US where numbers are growing rapidly, communities must understand the size and distribution of their homeless populations, as well as how information and resources are diffused throughout homeless communities. Currently, two sources of important but limited public data exist along such lines: count estimates of homeless across the US, gathered annually by the US Housing & Urban Development point in time (PiT) survey, and network information collected in a handful of small-scale qualitative studies of homeless social networks in various locations across the US (e.g. needle exchanges, information diffusion about the location of new shelters, locations of public restrooms). My research seeks to help communities leverage these types of publicly available data to better estimate/understand their own homeless populations and their impacts, in a cost-effective and reliable way. In this talk, I will introduce a method for using spatial Bernoulli graphs to simulate large-scale, spatially-embedded social networks, and show how one can perform a diffusion analysis over simulated homeless networks at the city-level. In particular, I will share examples of large-scale simulated networks of in-person and online (youth) homeless-to-homeless interactions in the City of San Francisco. From these examples, we will see much stronger spatial hopping and faster diffusion processes over the youth network, which indicate that information passing among homeless youth is faster and wider-spread than the in-person homeless network of all ages. Insights into the dynamics of homeless social networks have the potential to benefit public health efforts, such as the distribution of information , goods or services to homeless populations (e.g. where to report a fire or find an open public restroom). More generally, these simulation methods provide a unique way to visualize and characterize large-scale, hard-to-measure, spatially-embedded social networks and their interaction with the environment.