Machine-to-Machine (M2M) communications are a key enabler of Internet of Things (IoT) applications. Smartphones are a natural choice to serve as gateways (GWs) in mobile M2M communications due to their enhanced connectivity, powered by the ubiquity of mobile networks, and sensing capabilities. However, the use of smartphones as M2M GWs can have an impact on the smartphone usability and introduce undesirable battery depletion due to network accesses. In this work, we present a packet transmission scheduling model that exploits the concurrent use of multiple technologies in heterogeneous networks scenarios while guaranteeing time requirements. We model the system, including detailed network interface power consumption and latency, using linear programming (LP), thus building a useful tool for analysing and comparing network performance under different configurations. For a given set of packets to be transmitted and a given set of networks available, the model decides the best scheduling according to a cost function and ensuring that no deadline is missed. Our results show that a schedule minimising the energy consumption can reduce in 79% the total energy consumed when compared to a schedule that minimises delivery times, and thus allowing smartphones to achieve longer battery life. The trade-off is an increase in the total delivery time, but all deadlines are met.