"... In PAGE 10: ... 4. Results Table1 presents the average power and battery lifetime for the first-phase experiments.2 The benchmarks were run at a variety of frequencies and with the processor core at two different voltages.... In PAGE 10: ... Not surprisingly, the battery capacity drops slightly with the higher workloads. Table1 also shows the average power data taken for the corresponding experiments during the second phase. As ex- pected, this figure is always within the error on the average 2 When these results were first reported [19], the power presented did not include the power dissipated by the sense resistor (i.... In PAGE 10: ... Because the sense resistor used to measure the battery current is required by Itsy apos;s battery monitor, it should probably be con- sidered as part of the system. Therefore, the average power reported in Table1 is based on the same data, but has been re-calculated (as has its error) to account for the sense-resistor power. power during the battery-lifetime experiments, which con- firms that 3.... In PAGE 15: ...4 [19], the total power did not include the power dissipated by the sense resistor. As mentioned in Section 4, the average power P battavg reported in Table1 is based on the same data, but has been re-calculated to account for the sense-resistor power. This appendix presents the modifications made to the power calculation and to the error analysis, as well as the resulting values.... ..."

"... In PAGE 17: ... The worst case calculated error introduced by this measurement procedure is less than 9% for all of the experiments, 5% if deep sleep is excluded [12]. The power and battery lifetime data are presented in Table6 . Data for the intermittent-use scenario is given in rows 1 to 9, and the continuous-use scenario in rows 10 to 16.... ..."

"... In PAGE 17: ... The worst case calculated error introduced by this measurement procedure is less than 9% for all of the experiments, 5% if deep sleep is excluded [12]. The power and battery lifetime data are presented in Table6 . Data for the intermittent-use scenario is given in rows 1 to 9, and the continuous-use scenario in rows 10 to 16.... ..."

"... In PAGE 2: ... Table1 : Time required to double performance. There is a tremendous amount of work being done in the eld of microprocessor design, and many power management techniques are being built-in in the hardware of the micro- processor itself.... ..."

"... In PAGE 8: ...e.g., ad hoc routing or data fusion). Table4 shows the results of sharing the available currentcy be- tween the MP3 player (x11amp) and netscape for various amounts of available currentcy. X11amp requires 80mW to ensure contin- uous playback, the remaining capacity is available for netscape.... ..."