We have found strong excitement for many of the novel elements of the proposed WiNEST infrastructure. We reached out to many researchers with interests in next-gen wireless networking with an executive summary of WiNEST, asking them to identify topics that resonated with their own interests and for which no existing testbed offers similar access for experimentation. Several common threads emerged from the multiple responses. The researchers all agreed that a rural-centric testbed with a mix of new, dynamic access and wireless backhaul technologies inclusive of mobile nodes (whether vehicular or aerial) would be a significant novelty. Further, deployment of wide-area sensor networking to complement and expand core cellular network functions was also highly desirable. Through this survey, it became clear that WiNEST encompasses all these enablers and will indeed become a unique experimental wireless infrastructure to serve a broad research community. In what follows, we summarize the common set of topics that the respondents are most excited about.
Software- defined UAV networking
The programmable radio components, aviation control modules, and network stack on WiNEST’s UAS nodes will enable a new domain of research in software-defined UAS networking. The ability of UAS to maintain link quality over a mission is largely underexplored, requiring integration of both the communications and navigation functions. The on-board SDRs, autopilot software, and ground stations can be a key of research to ensure safe operations. Further, multiple UAS can operate simultaneously commanded by wirelessly connected hand-held devices (distributed ground control nodes). Such a setup presents new challenges vis-a-vis network connectivity and reliable and (power) efficient operation, likely requiring multi-objective design optimization.
Spectrum sensing and dynamic spectrum access in the wild
Several of the WiNEST use cases rely on spectrum sharing between unlicensed and licensed users. In turn, this will need reliable spectrum databases with real-time spectrum usage information. Based on prior team experience in architecting a passive, distributed network of spectrum monitors, WiNEST can dual-use several radio locations as spectrum sensors on both fixed terrestrial and airborne nodes. The SDR components will enable researchers to explore various cognitive networking mechanisms, and spectrum sensing approaches such as MIMO-based directional sensing. A significant outcome would be a publicly available spectrum usage data infrastructure, enabling additional knobs for network operation, e.g., source identification and localization.
Dynamic wireless backhaul and access
WiNEST’s wireless mesh backhaul, with 10+ Gbps bit-rate, 10+ km range, and capabilities to steer antennas and dynamically reconstruct topology, can potentially realize the vision of “dynamic wireless fiber backhaul”. The advantages of dynamic high capacity links extend to air-to-ground access with electronically steerable phased- arrays. This vision entails multiple research challenges.
All-weather wireless backhaul through hybrid FSO/microwave links and dynamic route reconstruction
Although mmWave bands suffer more from pathloss and weather disturbances compared with microwave bands, they possess more spectrum resources and higher capacity. A low-cost, constantly available wireless backhaul requires fundamental research that can strike a balance. In addition, new route reconstruction mechanisms are needed to reroute the traffic to detour regions under extreme weather conditions, taking into account real-time weather information and spatial correlation of weather impacts.
Dynamic air-to-ground wireless fiber access
The electronic steerability of mobile mmWave radios will enable high-capacity air-to-ground links for disaster recovery sites. The key research issues here involve: scalable interference management and spatial reuse among tens of densely deployed drone gateways with thousands of possible beam directions; efficient beam management when both the gateways and ground clients are mobile; integrated backhaul and access for air-to-ground network, etc.
Wide area networking and sensing
Although sensor networks have been extensively explored, the LoRa/NB-IoT low-power wide-area networking (LPWAN) technologies bring new challenges that need breakthrough research, and can enable exciting RF-based wide-area sensing applications.
Software-defined wide-area sensor networking
The LoRa/NB-IoT standards assume extremely low-duty cycle, low-rate applications (tens of bytes per minute per node). But wide- area sensor applications, with multi-kilometer coverage per link, may involve a large sensor population and create much higher traffic density, which overload the network especially at scale. New network architecture, scalable medium access protocols, and even physical layer innovations need to be explored. WiNEST’s wide-area connectivity module is fully software- defined, programmable from the PHY layer all the way up to the application layer. It can thus enable a new research agenda in software-defined wide-area sensor networking to meet the challenges. In addition, LoRa and NB-IoT are two competing technologies for LPWAN, but there is no openly available testbed to allow researchers to make a comparison.
Long-range, low-power RF sensing
Changes in the physical environment, such as humidity, fires, and road traffic can dramatically change the dielectric properties of physical environment and/or the signal propagation channel, which can potentially be sensed with LPWAN RF signals. The long transmission range enables sensing of a large area, but this also means far more channel variations, posing challenges to the robustness of sensing algorithms. Also, novel signal processing techniques such as channel stitching and multi-antenna (or multi- node) coordinated sensing are needed to improve the sensing resolution and accuracy.
Long-range, high-rate, low-power video surveillance
IoT devices at the edge are pushing the limits of wireless bandwidth (e.g. cameras monitoring of fruit ripening, pests, and livestock in farms), while demanding extremely low power. One interesting direction to resolve the dilemma is “offloading” of power-hungry elements across different devices deployed in a rural LPWAN network. For example, ongoing work from Ganesan is looking at how RF signal processing operations such as carrier generation, filtering, amplification, and modulation conversion can be offloaded from a sensor to a nearby device that may have access to power (e.g. a solar-powered node in a farm). Other recent work such as NetScatter has shown that concurrent backscatter transmissions are possible, thereby boosting the bandwidth to enable low-power video streaming.