Sparse Mobile Ad-Hoc Networks (henceforth referred to as SPMANETs)
are a relatively new class of networks. In certain specific contexts, they are also known as Delay/Disruption Tolerant Networks (DTNs) wherein sparseness and elay are particularly high. These networks are characterized by intermittent contacts between nodes. In other words, SPMANETs' links on an end-to-end path do not exist contemporaneously, and hence intermediate nodes may need to store, carry, and wait for opportunities to transfer data packets towards its destination. In this paper, we present a novel queueing-theoretic approach to the problem of computing the throughput capacity of SPMANETs employing a \textit{store, carry, and forward} paradigm for communication. We consider a network wherein $n$ relay nodes, in addition to a single source-destination pair, move according to a known mobility model that exhibits \textit{stationarity}. The relay nodes are assumed to have finite buffer sizes of $B$ packets each. The source sends data to the destination nodes by means of a two-hop relay protocol. We provide a novel Markov-chain-based analysis for computing the throughput of such a network, resembling the approach used in Queueing Theory. We further provide a novel approach for state-space reduction in order to incorporate complexities arising as a result of random mobility, and practical considerations such as interference and contention. Finally, we validate our approach using simulations for two popular cases of mobility: random-walk-on-grid and random-waypoint mobility models.