# Analytics Capacity: Theoretical Formulation and Scaling Laws

 Abstract We present a theoretical framework for charac- terizing analytics capacity at the tactical edge. We consider a system with multiple edge nodes that are capable of running analytics applications. Each user runs its analytics instance in its nearby edge node. There are multiple types of analytics applications and all instances of the same application (serving different users) share the same storage. An additional amount of computation resource is consumed by every instance, which is not shared with other instances. Each edge node has a limit on the maximum number of applications it can host and the maximum number of users it can support. The analytics capacity is defined as the maximum number of users the entire system can support, where we assume that the user to edge node association is independent and identically distributed (i.i.d.) and each user runs one analytics instance. In this paper, we derive an upper bound and an achievable lower bound of the analytics capacity as defined above. Both exact and asymptotic expressions of the capacity bounds are given and their insights are discussed. Authors Shiqiang Wang (IBM US) Ting He (PSU) Prithwish Basu (BBN) Theodoros Salonidis (IBM US) Kevin Chan (ARL) Date Sep-2017 Venue 1st Annual Fall Meeting of the DAIS ITA, 2017