Bringing Cognitive Radio Into a New Dimension

"Capabilities that may have taken an entire team to develop in a lower-level language became feasible for two graduate students using the NI integrated hardware-software approach."

- Brian Beck, Georgia Tech Research Institute

The Challenge:

Building a testbed system that uses ultra-wideband (UWB) radio and software defined radio (SDR) to improve RF data collection by adding a spatial context to a cognitive radio system.

The Solution:

Developing a networked system of mobile robots that combines the spectrum-mapping abilities of cognitive radio (CR) with recently pioneered localization and spatial-awareness capabilities. By pairing a UWB and an SDR, such as the NI USRP-2922 transceiver, with positioning algorithms programmed using LabVIEW system design software, the cognitive spectrum operations testbed (CSOT) demonstrates a significant advancement in cognitive radio research.


A Paradigm Shift in Cognitive Radio

As communication systems grow in complexity and number, it is increasingly important that devices independently optimize spectrum use to ensure reliable communication channels. The software defined radio lab at the Georgia Tech Research Institute explores a variety of signal-processing and CR research areas. In CR systems, individual software defined devices can detect available spectrum and tune parameters accordingly to suit the environment. To address the need for improved RF data collection for numerous applications, our lab is pioneering the concept of spatial CR. In this paradigm, the aforementioned SDR capabilities combine with spatial awareness: localization, spectrum mapping, physical area mapping, and mobility. Through software defined algorithms, network nodes become aware of their physical surroundings—they can adapt and reposition for improved throughput, spectrum utilization, and security.


An Integrated Hardware-Software Approach

We developed a networked system of mobile robots with onboard processing and communication capabilities. Each unit is mounted on the iRobot Create platform, which provides standard mobility and collects data from the bump sensor and encoder wheel. An NI USRP™ (Universal Software Radio Peripheral) and UWB transceiver provide SDR and spectral-analysis capabilities. Finally, a standard laptop houses the controller, which collects measurement data, issues commands to the iRobot, and runs the SDR client code programmed using LabVIEW.


For software development, we chose LabVIEW because of its ability to seamlessly interface with our hardware at a high level without sacrificing full design control. LabVIEW supports the iRobot platform and both radio transceivers through drivers and APIs, which simplified connectivity and device communication. CR and spatial awareness algorithm development and testing was straightforward and intuitive to develop using the graphical environment. Finally, when implementing the client-server model, by using high-level LabVIEW TCP/IP functions, we could easily build a scalable and reliable mechanism to stream data from anchor-free nodes to a central server.


Rapid Prototyping for Success

The end result was a flexible testbed powered by the NI platform. By simplifying the controller and SDR components, we could efficiently develop and test our spatial-mapping algorithms for rapid prototyping. Capabilities that may have taken an entire team to develop in a lower-level language became feasible for two graduate students using the NI integrated hardware-software approach. While exploring new research, it was also extremely helpful to have the support of dedicated NI sales and support personnel, who were instrumental in reducing initial barriers to success.


The individual nodes are tracked using UWB ranges, odometry, and multidimensional scaling[2][3]. After the client processes the data, the algorithm creates a continuous spatial map of the average power for the surrounding area. This spatial awareness could be used for a variety of optimization methods, including positioning a relay node to increase communication reliability.


In demonstrating CSOT capabilities (Figure 1), the spatial CR derived localization of each individual node with impressive accuracy. In this experiment, we collected the UWB range and encoder wheel data from the mobile nodes as they moved down a hallway. Using the algorithm developed in the prototyping stage, we calculated the location over time and sent it to the central server[2]. As shown to the left of the image in Figure 1, the estimated node positions are plotted with the surveyed positions to validate the design. The estimation had an average error rate of less than 2 cm.


Less Time Coding and More Time Testing

In this CR field, mapping spatial context to spectral data is a significant and exciting research area to explore. By using LabVIEW to streamline device connectivity, controller design, client/server management, and algorithm development, we could spend less time coding, and more time testing. With this unique NI hardware-software approach, we pursued new concepts and accelerated CR system discoveries.



  1. R. Baxley, B. Beck, J. Kim, and B. Walkenhorst, "RadioBOT: A Spatial Cognitive Radio Testbed," (presented at IEEE Aerospace Conference, 2013).
  1. R. Baxley and B. Beck, "Anchor-Free Node Tracking Using Ranges, Odometry, and Multidimensional Scaling," (presented at IEEE International Conference Acoustics, Speech, and Signal Processing, 2014). 
  1. R. Baxley, B. Beck, and J. Kim, "Real-Time, Anchor-Free Node Tracking Using Ultra-Wideband Range and Odometry Data,"  (presented at IEEE International Conference on Ultra-Wideband, 2014).



Author Information:

Brian Beck
Georgia Tech Research Institute

Figure 1. The CSOT performing a localization demo. The estimated vs. surveyed position is plotted on the left.