The conventional architecture for short-range, low-power, wireless hardware integrates a radio transceiver and a general-purpose microcontroller. However, microcontrollers have limited computational power and hence, restrict the kinds of algorithms that can be implemented. Moreover, closed radio architectures prevent the direct access to the physical layer needed for many novel applications. The emergence of low-power FPGAs with their efficient duty cycling support presents an opportunity to create a novel wireless platform to overcome these limitations. The project develops a highly flexible hardware/software architecture for mobile wireless networking based on low-power FPGA devices. High performance DSP and other algorithms can be implemented directly in hardware, while the rest of the code runs on a soft processor core inside the FPGA. This flexible hardware/software boundary increases the complexity of software development that is eased via sophisticated development tool support and an extensive software component library that the project is also developing. The small inexpensive platform offers two orders of magnitude higher performance than current microcontroller-based hardware at equivalent power for many applications. Hence, it will stimulate research in novel protocols and enable new applications not possible today. A competition for undergraduate students will be organized in which winners will receive the hardware, software, training, and mentorship needed to carry out their ideas during a paid summer internship at Vanderbilt.
Software-defined radios are reconfigurable communication systems that transcend historical boundaries between hardware and software subsystems, physical and logical layers, and analog and digital domains. In so doing, they enable radical new architectures, novel radio designs, and high-performance wireless protocols that are not easy to design, implement, or evaluate using traditionally-layered approaches that rigidly partition functionality. Although modern SDR platforms have been used to explore many facets of the wireless design space, their typical architecture makes it very difficult to explore the small, inexpensive, and low-power design space. As a result, important application domains like mobile phones, sensor networks, visible light communications, and radio frequency localization--that could benefit from radical approaches, but which require small form factors, low unit costs, or low-power operation--remain relatively unexplored. This project demonstrates that a software radio with an index card form factor that costs $150 and offers smartphones battery life from just a pack of 'AA 'batteries is possible, and that it enables new research. The intellectual merit of this work lies in a hardware and software platform that allows flexible partitioning of applications across system resources--either by time-multiplexing computations on shared resources like a central processing unit, or parallelizing those computations on dedicated resources like an underlying FPGA fabric--using a software-radio and its applications as the motivating testbed. Several applications are themselves novel contributions, including A-MAC (a receiver-initiated wireless protocol that offered best-in-class performance) and Harmonia (an RF TDoA localization system). The broader impacts of this work stem from its enabling nature for many research topics in low-power wireless networking including protocol design (e.g. receiver-initiated protocols, concurrent transmissions, efficient flooding), software-defined lighting applications (e.g. lighting, energy efficiency, visible light communications, visible light positioning, optical time synchronization, and others), and RF-based localization (e.g. tracking the position of quadrotors indoors at high update rates). The ideas pioneered in the program have led to new commerical SDR systems and improvement to FPGA technology to better support low-power SDR, and have launched or accelerated new academic research areas. Key among these include receiver-initiated communications, concurrent transmissions, and software-defined lighting. The program has also helped train multiple graduate students. Its artifacts have been adopted for undergraduate computer engineering courses and its intellectual property has been open-sourced for third parties to freely use and build upon.
