Worldwide, mobile data connectivity is now widespread, but not yet ubiquitous due to coverage limits and cost concerns. Mobile data ofioading to WiFi-where available- could greatly decrease the usage of cellular data networks. In delay-tolerant applications, one could delay network communication in order to exploit free WiFi connections expected to appear soon. However, WiFi connectivity is limited, and even delay-tolerant applications must meet qualityof-service deadlines. To explore such bandwidth scheduling issues, we develop an optimal MILP-based scheduling framework. Our framework schedules multiple application data streams with varying size and delay tolerance, onto networks with varying coverage and bandwidth, in order to minimize cellular data usage. The ability to subdivide data streams into scheduling units is important, because it allows applications to exploit brief windows of WiFi coverage and it allows tradeofis between solution quality and solver runtime.