Current Projects

Classical, model-based techniques for resource management and coordination in wireless networks often rely on a priori assumptions that are too idealized for real deployment scenarios. These techniques generally fail to capture the broad range of parameters and channel access controls that impact network performance, especially in situations involving coexistence of heterogeneous technologies on a shared spectrum. Machine learning (ML) techniques, on the other hand, enable the adaptation of a wide range of control parameters (features), with no or minimal assumptions on the underlying channel and network dynamics. ML techniques have recently been studied and applied to numerous wireless systems, from sensor networks to 5G millimeter-wave (mmWave) transmissions. The stringent quality of service (QoS) requirements associated with certain use cases of 5G+ wireless systems (e.g., ultra-reliable low-latency communications and eMBB) along with the integration of edge computing (MEC) platforms into these systems have triggered intensive research on the use of ML for wireless communications. 

The overarching goal of our research endeavor is to explore ML techniques for fast protocol adaptation, resource allocation, and spectrum sharing in both mid-band (sub-6 GHz) as well as high-band (mmWave) wireless systems. Our research strategy is to combine model-free ML tools (“black box”) with robust and highly generic modeling assumptions (hence, a “grey box” overall design) so as to achieve computationally efficient real-time adaptation of complex wireless networks under limited information. We leverage the flexibility of ML techniques to incorporate a large set of diverse but observable inputs into the learning engine. In contrast to existing state-of-the-art ML-based techniques that over-emphasize performance over complexity, our proposed designs target fast/real-time adaptation. Specific themes in our ML-based research agenda include, but are not limited to:

• User Tracking and Link/Beamwidth Adaptation in 5G mmWave Systems 
• Interference Mitigation in Contention-based Wireless Networks 
• Intelligent Scheduling of Time/Frequency Resources in OFDMA Wireless Systems
• Misbehavior Detection and Protocol Compliance for Unlicensed Wireless Networks
• Lightweight Machine Learning for Efficient Frequency-Offset-Aware Demodulation

Over the last two decades, dynamic and opportunistic spectrum access (DSA/OSA) has been advocated as a new paradigm for improving the utilization of the licensed spectrum below 6 GHz. Spectrum regulators (e.g., FCC, NTIA, etc.) embarked on numerous initiatives that aim at identifying underutilized portions of the licensed spectrum and exploring new models for spectrum sharing among different types of users. These initiatives include the FCC's ruling on TV broadcast channels in the 54 to 698 MHz range (so-called TV whitespaces or TVWS), the FCC's Notice of Proposed Rule Making (NPRM) on the 3.55-3.65 GHz radar band (which lays the groundwork for coexistence between small-cell LTE systems, military radar, and other incumbent systems), and others. Similar initiatives are taking place in other regions of the world, such as the European Union (EU)'s efforts to explore the 2.2-2.3 GHz band for spectrum sharing between aeronautical telemetry radar and small-cell LTE systems. The role of radio spectrum as a critical economic growth engine was highlighted in the 2012 President's Council of Advisors on Science and Technology (PCAST) report, which recommended creating "the first shared-use spectrum superhighways."

Yet, despite this strong push, commercial DSA wireless services have not proliferated. This is in part due to the limited availability of TVWS in urban areas and, more critically, their spatial non-uniformity across the Continental US. Broadcast TV stations in the US are licensed to operate over 6 MHz wide (NTSC) channels that fall into four noncontiguous segments of the spectrum (52-72 MHz, 76-88 MHz, 174-216 MHz, and 470-698 MHz). In a given urban location, the number of TV channels available for opportunistic access can be as low as 2 to 4, after accounting for FCC protection rules (e.g., power masks, adjacent-channel interference, etc.) and "TV pollution" zones, i.e., areas where TV signals are not decodable but are strong enough to disrupt secondary (opportunistic) transmissions. Not only that these channels are too few, but they often have to be shared by several secondary users (SUs), raising the prospects of SU-SU interference. The non-contiguity of available TV channels further complicates matters, as complex carrier/channel aggregation techniques must now be used over a given link to achieve WiFi-comparable channel widths (e.g., 20 MHz and higher).

The overarching goal of the underlying project is to enable DSA operation over a large swath of the opportunistic spectrum below 6 GHz, including the UHF portion of TVWS (470-698 MHz segment), the 3.5 GHz radar band, as well as other bands in the 1 to 4.4 GHz range that may later become candidates for DSA. By aggregating different (not necessarily contiguous) portions of the RF spectrum at the same wireless device, we anticipate that variations in white/grey spaces can be significantly reduced, allowing for sustained DSA operation over a large geographical area. Enabling such operation while maintaining high link throughput requires advances in the following areas:

• Electrically small tunable antennas that allow for dynamic access to narrowband (e.g., 6 MHz) channels, but yet support a broad frequency range from 470 MHz to 4.4 GHz.
• Channel/carrier aggregation techniques for noncontiguous narrowband channels.
• Multi-channel MIMO functionality that allows a given secondary link to boost its throughput (via spatial multiplexing gain) and, simultaneously, minimize interference onto other coexisting systems (via MIMO precoding techniques).

The first capability is intended to beef up the agility of the secondary radio, allowing it (for example) to switch between TVWS and the radar band, on demand. The last two capabilities are intended to significantly increase the per-link throughput and reduce SU-SU interference. Through novel advances in the above areas, we plan to design, implement, and experimentally evaluate a multi-channel 2-by-2 MIMO system for DSA with a broad RF coverage. Our research agenda involves optimizations at both the signal level (via MIMO precoders) as well as the antenna level (via highly tunable small form-factor reconfigurable antennas).

Control dimensions in a CMIMO WLAN: (a) Frequency allocation among multiple links, (b) power allocation over frequencies and antennas, and (c) beamforming-based interference suppression. The example shows a WLAN of 4 links, 2 antennas/node, and 3 channels.

Common handheld wireless radios operate in half-duplex (HD) mode, with one transceiver alternating between transmission and reception. Full-duplex (FD) operation in which a device is capable of simultaneous transmission and reception (STAR) can be achieved by employing multiple transceivers at the same device or (more recently) through orthogonal frequency division multiple access (OFDMA) technology. In both cases, however, STAR is performed over different transmit and receive channels (henceforth called STAR-D capability). Sufficient separation (i.e., guard bands) is needed between the transmit and receive channels to prevent side-lobe interference and spectrum leakage from corrupting the received signal. Until recently, performing STAR over the same channel (a mode that we refer to as STAR-S) was deemed impossible due to strong self-interference. This common belief was recently challenged by several research groups, which have successfully demonstrated the possibility of FD communications using a combination of self-interference suppression (SIS) techniques, including RF analog cancellation, digital baseband interference cancellation, circulators, phase shifters, etc. In fact, it has been shown that a node's own transmission can be suppressed by 73 to 110 dB (noise floor), depending on the underlying SIS schemes. Such a level of SIS has significant ramifications on network protocols, which are often designed with the assumption of HD operation. In this project, we aim at exploiting SIS capabilities in designing efficient protocols for wireless networks. Two main thrusts are being pursued:

• Exploiting SIS in Dynamic Spectrum Access Systems
• Designing Adaptive SIS-aware Multi-channel MAC Protocols for OFDMA-based Wireless Networks
• Studying the Feasibility of Full-duplex Communications from Network's Perspective

Establishing communications in a multi-channel wireless network requires the communicating devices to "rendezvous" on a common channel before carrying out normal data transmissions. The rendezvous process enables the exchange and negotiation of critical information, including transmission parameters, connectivity and topological changes, etc. This process takes place at the time a new session is established and following the loss of connectivity between the transmitter and receiver (e.g., due to jamming and/or channel fading). Clock drifts make it difficult for the communicating parties to maintain a common time reference, so the rendezvous process must be robust to misalignment and lack of tight synchronization. Moreover, in a dynamic spectrum access (DSA) environment, rendezvousing may need to be conducted under a heterogeneous opportunistic spectrum setting, i.e., the set of idle channels are generally different for different nodes. This heterogeneity is caused by spatiotemporal variations in spectrum availability as well as interference/jamming conditions. In this project, we are developing novel quorum-based frequency hopping (FH) schemes for unicast and multicast rendezvous in wireless networks. The quorum approach makes it possible to come up with more structured (less random) FH sequences, hence increasing the likelihood of overlap (rendezvous) between two or more nodes. At the same time, this structured approach makes devices inherently vulnerable to "insider attacks," i.e., attacks by compromised nodes that exploit knowledge of the underlying quorum system to prevent the rendezvous process for succeeding. Our research deals with both the performance (speed of rendezvous) as well as security (vulnerability to adversarial attacks) issues of quorum-based FH rendezvous designs, under various spectrum and adversarial settings.