Low power radios enable ubiquitous sensor networks that can be used for a variety of applications, such as biomedical monitoring, environmental sensing and intrusion detection. However, existing transceivers have been unable to demonstrate low enough power consumption to fully realize these applications. Low duty cycle impulse radio can offer significant power savings by allowing the transceiver to be turned off between bits. Ensuring that two nodes are sufficiently synchronized to duty-cycle in this fashion is a significant challenge. To solve this problem we used pulse coupled oscillator (PCO) scheme of Mirollo and Strogatz. We performed extensive simulation of the PCO network under a realistic radio parameter space and found synchronization to be robust. We then implemented a low power, aggressively duty-cycled dual-band IR-UWB transceiver in an IBM 90nm CMOS process based on this synchronization mechanism. The transceiver features an energy-detecting front-end, a relaxation oscillator based PCO and a precise edge locking PLL for time bin generation. The time-bins provide our system with 123 unique channels that can be used for multiple access.. We constructed a FPGA based test and measurement setup and implemented a synchronization management finite-state-machine on microprocessor. The PCO network is shown to synchronize nodes robustly with experiments confirming the results of our simulations. We found that the synchronization management scheme allows a four-node system to remain synchronized with duration-of-synchronization iii on the order of one second when using 30ns RF-on time windows in a 7.2 frame. As a result of this aggressive-duty cycle of 0.8%, the total transceiver power consumption is reduced to 119uW while actively communicating. We were able to demonstrate functional radio links transmitting packets of upto 1200bit length over a meter range, proving the viability of the concept. Finally we perform an analysis of a simplified theoretical model of the system which provides fundamental limits to the size of the network that can be supported and the data throughput that can be achieved. The analysis shows that the scheme offers significant power savings benefits for up-to ten nodes if bit-error-rate can be sufficiently controlled. iv