An overwhelming and rapidly-increasing number of technologies that drive economic growth and humanity’s well-being hinge on wireless connectivity such as Wi-Fi, cellular networks and Bluetooth. Yet, only 8% of the radio spectrum is allocated to cellular, Wi-Fi and Bluetooth networks combined. Thus, a critical need emerges for mechanisms that help us understand and control our spectrum use more efficiently. This project develops algorithms and systems that jointly consider the physical and fucntional components of spectrum measurement to enable emerging dynamic spectrum applications.
The process of spectrum measurement involves three key steps: (i) spectrum sensing, (ii) data management and (iii) spectrum characterization. The goal of this project is to design algorithms for lightweight, real-time and unsupervised collection (AirSWEEP), management (AirPRESS) and characterization (AirVIEW) of spectrum data to improve the learning outcomes of spectrum sensing. Our work bridges the gap between spectrum algorithms and infrastructures by formalizing and evaluating the effects of heterogeneous infrastructures on spectrum learning outcomes.
TxMiner is a spectrum analytics engine, for unsupervised detection of arbitrary transmitters without prior knowledge of transmitter characteristics. TxMiner leverages the phenomenon that fading of non-line-of-sight wireless signals follows a Rayleigh distribution, while noise follows a Gaussian distribution. Thus, the raw spectrum samples can be modeled as a mixture of Rayleigh and Gaussian distributions. Based on this observation we design a machine learning algorithm that extracts Rayleigh and Gaussian sub-populations from a given RF signal population.
Next-generation wireless network design should happen with an outlook towards connecting the “last billion”. Our work at UbiNET Lab has analysed off-the-shelf mobile wireless technologies in various infrastructure-challenged environments from rural Africa, trough refugee camps and agricultural lands in the U.S. Our findings inform our research on wireless network and protocol design, integration and in-situ deployment.
This project designs a socio-technical framework in support of rural emergency preparedness and response. First, the project will develop a heterogeneous network architecture and corresponding protocols that leverage wide-area wireless backhaul over TV white spaces, WiFi and pocket-switching to provide (i) continuous communication to first responders and (ii) delay-tolerant information access to residents. In addition, the project will design a smartphone app which will support the collection of information from different sources and its exchange among first responders, government agencies and residents. Second, the mobility patterns and network availability collected will enable the development of a dynamic probabilistic community network model. Novel graph-theoretic algorithms will identify information-depleted sub-communities and inform optimal information dissemination strategies. Finally, the project will assess adoption and use of the technologies by various community members to maximize the benefits associated with timely, rich and high-quality information, disseminated through technological devices. The framework will be developed in collaboration with the Town of Thurman, NY and the Emergency Service Department in Warren County, NY.
The goal of this project is to design, develop, integrate and deploy an end-to-end system for real-time agricultural data collection, analytics and control. To this end, we employ fundamental knowledge in wide-area wireless networks, digital signal processing, machine learning and control, to design (i) robust control mechanisms for multi-sensor agricultural data collection and fusion, (ii) wide-area, heterogeneous wireless networks for ubiquitous farm connectivity, (iii) algorithms and models for farm data analytics that produce actionable information from the collected agricultural data, and (iv) novel control mechanisms forautonomous, proactive farming.
HybridCell allows coexistence and simultaneous use of commercial and locally-owned cellular networks. It enables low-cost services for local users and organizations, and can also complement commercial cellular networks’ functionality in rural areas by providing mobile data services.
We utilize large-scale real-world traces from cellular and Internet access networks in order to understand network performance and user behavior. We design solutions after careful consideration of actual communication needs in rural communities.
Kwiizya is a cellphone network technology that provides basic voice and text messaging in remote rural areas for free. In the summer of 2012 we deployed an instance of Kwiizya in the remote village of Macha, Zambia.
We develop long distance wireless solutions that utilize white spaces to bring connectivity to and within remote communities. My focus is on resource allocation that is informed by the channel quality as perceived by the communicating parties.
AirLab is designed to be a publicly accessible distributed infrastructure for wireless measurements. It will facilitate meaningful analysis of wireless networks and protocols, by providing consistent and comparable wireless traces. AirLab consists of multiple homogeneous measurement nodes deployed across various sites around the globe. Access to these measurement nodes is controlled by a control center, AirLab Central, located at UCSB. Researchers can schedule experiments on the measurement nodes, which would allow them to gather specific traces from the host networks. All the traces are uploaded to a central repository and are available to the public after sufficient anonymization to protect privacy of the site involved.