battery with a specific energy exceeding 300 Wh/kg would exceed the performance of both DMFC and Li/MnO2 and would be a good candidate.
As with the 20-W regime, if the same battery were used for average and peak power, the degree to which the peak power demand would degrade the specific energy would depend on the duty cycle. In the 2-W regime, a capacitor could be used for the 5-W peak if the pulse width is sufficiently short. An appropriately sized Li ion battery or a capacitor would most likely provide the 5-W peak power for the passive DMFC option, which would be optimized for the average power load.
At such low powers, advances in thermoelectric materials make them viable candidates for energy conversion systems. Thin film materials could enable very lightweight thermoelectric modules. Furthermore, they could be used in conjunction with catalytic combustors employing jet propellant 8 (JP-8) fuel, which has twice the energy content of methanol. However, even extremely optimistic energy efficiencies for a thermoelectric system are less than the 20 percent efficiency of a passive DMFC system. Other factors, such as system simplicity or size and weight, could make the thermoelectric option viable for some applications.
At the 2-W power level it is also possible that energy harvesting technologies could affect the overall weight and volume of the power source system. As with a DMFC system, the total energy harvesting system would have to perform significantly better than the battery it would replace. The total system weight for a given power produced and the often-intermittent nature of energy harvesting technology must be taken into account in any trade-off analysis. The significant benefit of using energy harvesting is the inexhaustible supply. This would free future warriors from concern about having no backup for dead batteries.
A 2-W soldier system would require a system-level approach to design that would consider both the energy consumers (sinks) and the power sources. Many techniques can be used to improve the energy-efficiency of a system, from the network level down to the physical level of the battery. At the network level, routing methods tailored to the power demanded by the network subsystem can improve power levels by 15 percent on average and reduce latency by 75 percent relative to methods that consider only the transmitted power.
At the boundary between the network and the processor levels, a computation can be performed locally or remotely depending on the relative performance of the local and remote system, the transmission bandwidth and power demand, and the network congestion. The largest power demand in a mobile computing system is for communication and computation. Techniques for reducing the energy usage in these areas include energy-aware network routing, balancing local and remote processing, and changing the central processing unit (CPU) speed dynamically.
Techniques for communication and computation cannot be studied in isolation: one technique for reducing communication energy usage is to perform more local processing, but this increases the amount of computation energy. Thus in the case of local vs. remote processing, the communication and computation subsystems must be considered together. For the system studied by Martin et al. (2003), energy can be conserved by remote processing any task that requires more than 1.4 milliseconds of processor time per kilobyte transferred. At the processor level, the main memory bandwidth has a significant effect on the relationship between performance and CPU frequency, which in turn determines the energy savings of dynamic CPU speed-setting.