Collaborative Research: Practical Foundations for Networking with Many-Antenna Base Stations

NeTS Large


This project is a collaborative endeavor funded by the NSF NeTS program. It involves Rice University and Ohio State University. It leverages the Argovs many-antenna MU-MIMO base-station testbed built and hosted at Rice University.

Project Team

Principal Investigator and Co-Principal Investators

  1. Lin Zhong (PI)
  2. Edward W Knightly (Co-PI)
  3. Ashutosh Sabharwal (Co-PI)
  4. Ness Shroff (Co-PI) (Ohio State)

Graduate Students and Postdocs

  • Jiayu Pan (Ohio State)
  • Guidan Yao (Ohio State)

Collaborators

  • Kannan Srinivasan (OSU)
  • Morteza Hashemi (University of Kansas)
  • Yin Sun (Auburn University)
  • Bo Ji (Virginia Tech)
  • Jia (Kevin) Liu (OSU)

Alumni

  • Morteza Hashemi (Postdoc, Ohio State)
  • Zhenzhi Qian (Ohio State)
  • Fei Wu (Ohio State)
  • Ming Zhang (Ohio State)
  • Clay Shepard
  • Abeer Javed
  • Evan Everett
  • Leo Meister (REU)
  • Andrew Brooks (REU)
  • Michael Tsehaie (REU)
  • Yin Sun (Postdoc, Ohio State)

Undergraduate Students

  • Bowen Liu
  • Andrew Pham
  • Aryan Sefidi

Project Goals

The goal of the proposed project is to provide the much needed practical foundations for networking with many-antenna base stations. In the project, we will not only explore novel approaches toward scaling up the number of base-station antennas to 10s and even 100s; but also rethink the entire network architecture exploiting the emergent properties as the number of base-station antennas grows large. Our approach is a combination of theory-driven simplification, measurement, and testbed-based implementation. In particular, the project will leverage ArgosNet, a multi-cell reconfigurable wireless network testbed of many-antenna base-stations recently funded by an NSF CRI grant with up to 400 antennas per base station. With many-antenna base stations deployed on rooftops and indoors on Rice campus and fully mobile terminals, the testbed provides unprecedented opportunities to study many-antenna MIMO systems in real world.

Toward providing the practical foundations for many-antenna MIMO networks, the project targets at three interrelated areas of innovations. (i) Scalable Control and Coordination. The project will develop scalable designs of control and coordination functions for MU-MIMO networks. In particular, it will design a suite of protocols for network-wide CSI collection that significantly reduce its overhead.

(ii) Scalable Resource Allocation. The project will develop novel resource allocation solutions for many-antenna MIMO networks in order to support both high data rates and low-latency requirements. It will contribute a novel scalable scheduling framework that use slow-time-scale information or statistical channel information and design scheduling policies based on MIMO rateless codes.

(iii) Empirical Foundations from Measurements. The project will perform previously impossible real-time measurements of MU-MIMO channels in order to understand channel correlation, variation and reciprocity and their relationships with spectrum band, mobility, and hardware impairment. In particular, the project will derive novel models, reciprocity calibration methods, and novel channel state representations that will power research in network designs.


Accomplishments

Major Activities


In YEAR ONE

Large-Scale MIMO Channel Measurement and Analysis

To better understand these MU-MIMO channels in the real-world, we built a realtime wideband many-antenna MU- MIMO channel measurement system that supports high time- frequency resolution across the UHF, 2.4 GHz, and 5 GHz bands. We built this system on the ArgosV2 platform and leveraged the Faros control channel design to provide time-frequency synchronization and Channel State Information (CSI) collection. To support UHF, we ported Argos and Faros to the WURC platform. We performed an extensive measurement campaign that includes fully mobile traces across the UHF, 2.4 GHz, and 5 GHz bands in diverse environments. At 2.4 and 5 GHz, we collected traces with up to 104 base station antennas serving 8 users in both indoor and outdoor environments, with varying mobility. At UHF, we collected traces with up to 8 base station antennas serving 6 users in both indoor and outdoor environments, with varying mobility. These traces typically have frame lengths, i.e. time resolution, varying from 2 ms to 40 ms. These traces allowed us to investigate large-scale MIMO channels. We are in the process to make them openly available via argos.rice.edu.

Angular Domain Representation of Large-Scale MIMO Channel

A key challenge in large-scale MIMO is the large overhead in CSI acquisition. We designed two novel types of angle-of-arrival (AoA) based beamforming schemes that harness the reciprocity of dominant AoA. Both schemes require CSI acquisition overhead that only scales with the number of served mobiles, not the number of base-station antennas. We analyze the performance of the proposed schemes both analytically and numerically. We show that both our proposed schemes lead to sum throughput that scales with the number of base-station antennas, and hybrid beamforming performs close to ideal zero-forcing beamforming.

Mobility-aware MAC Protocol Design

Performance of many-antenna basestations is highly sensitive to client mobility. For example, large-scale MIMO link can lead to high directivity. As a result, they must address new link training and adaptation challenges due to both client and environmental mobility. We designed, implemented and evaluated MOCA, a protocol for Mobility resilience and Overhead Constrained Adaptation for high directivity links. The team introduced Beam Sounding as a mechanism invoked before each data transmission to estimate the link quality for MIMO precoding designs, and identify and adapt to link impairments. We devised proactive techniques to restore broken directional links with low overhead and design a mechanism to jointly adapt beamwidth and data rate, targeting throughput maximization that incorporates data rate, overhead for CSI collection, and mobility resilience.

Scalable MU-MIMO Uplink for Many-Antenna Base Stations

Mobile devices have fewer antennas than APs due to size and energy constraints. This antenna asymmetry restricts uplink capacity to the client antenna array size rather than the AP’s. To overcome antenna asymmetry, multiple clients can be grouped into a simultaneous multi-user transmission to achieve a full rank transmission that matches the number of antennas at the AP. In our work, we design, implement, and experimentally evaluate, the first distributed and scalable system to achieve full-rank uplink multi-user capacity without control signaling for channel estimation, channel reporting, or user selection.

Limited Channel State Information on Massive MIMO Network Performance

In recent years, there have been significant efforts on the research and development of Massive MIMO (M-MIMO) technologies at the physical layer. So far, however, the understanding of how M-MIMO could affect the performance of network control and optimization algorithms remains rather limited. We analyze the performance of the queue-length-based joint congestion control and scheduling framework (QCS) over M-MIMO cellular networks with limited channel state information (CSI).


In YEAR TWO:

Understanding Real Many-Antenna MU-MIMO Channels

To better understand these MU-MIMO channels in the realworld, we implemented a realtime wideband many-antenna MU-MIMO channel measurement system. Built on the ArgosV2 platform and Faros control channel design, this system enables very reliable high time-frequency resolution measurements, supporting sub-millisecond sounding intervals with 20 MHz bandwidth, across the UHF, 2.4 GHz, and 5 GHz bands. We leveraged this system to conduct one of the most extensive and diverse mobile MU-MIMO measurements campaign ever reported, already containing over 1 billion channel measurements on more than 20 topologies, and continuing to expand. These topologies include line-of-sight (LOS) and nonline-of-sight (NLOS) scenarios in both indoor and outdoor environments with various degrees of mobility and multipath. Additionally we constructed an open-source Python channel analysis toolbox to study the fundamental properties of manyantenna MU-MIMO channels.

Modelling and performance evaluation of many-antenna MU-MIMO

MU-MIMO boosts the capacity of wireless LANs by transmitting data streams to multiple users concurrently, thus scaling up the achievable data rate by a factor equal to the number of antennas on the base station. However, to observe the performance benefits of MU-MIMO in practice, these gains must propagate to the transport layer. Since performing multi-user downlink transmissions incurs overhead in CSI acquisition to transmit concurrent data streams, packet level characterization and queuing dynamics introduced by traffic flows for different users in the network determine the performance of the system. Thus, it is important to understand how microscopic packet level dynamics introduced by dominant transport layer protocols such as TCP affect the system performance. However, theory from available literature is insufficient to explain this impact. We have developed system models that make it possible to capture such microscopic queuing dynamics and its impact on the system performance. We have designed experimental frameworks to rigorously validate our model. Our analytical model, while advancing the theory on MU-MIMO, highlights important system parameters that need to be considered in the design of efficient and scalable protocols for deriving maximum performance gains from MU-MIMO at the transport layer.

Scalable Random Access for Massive MIMO

Random access is a crucial building block for nearly all wireless networks, and impacts both the overall spectral efficiency and latency in communication. In massive MIMO, many more users can be supported in the same time-frequency slots and that means, more users will have to contend for access. This in turn will increase the access delay and reduce overall spectral efficiency gains from massive MIMO.

Directional Training for FDD massive MIMO

Massive multi-input multi-output (MIMO), where the base station is equipped with a large number of antennas, can improve the spectral efficiency manifold. To leverage the full array gains, full channel state information at the transmitter (CSIT) is essential in massive MIMO. In the time-division duplexing (TDD) mode, uplink/downlink channel reciprocity can be leveraged to obtain downlink channel state information at the base station by using a small number of training pilots from mobiles via uplink. However, a significant fraction of spectrum allocations worldwide are for frequency-division duplex (FDD) operation. In FDD, channel reciprocity does not hold, which in turn has made FDD Massive MIMO a “grand challenge” problem.

Limited Channel State Information on Massive MIMO Network Performance

In recent years, there have been significant efforts on the research and development of Massive MIMO (M-MIMO) technologies at the physical layer. So far, however, the understanding of how M-MIMO could affect the performance of network control and optimization algorithms remains rather limited. In this woprk, we focus on analyzing the performance of the queue-length-based joint congestion control and scheduling framework (QCS) over M-MIMO cellular networks with limited channel state information (CSI).


In YEAR THREE:

Channel correlation in large antenna arrays

We have found channel correlation highly depends on antenna element placement, environment, and even users’ positions, and therefore can be different in different scenarios. In this project, we measured channels from a 96­antenna BS to 8 same­ radius single­antenna users in indoor environment with a LOS component. The measurements were taken in 2.4 GHz band over 20 MHz bandwidth using an 8x12 rectangular antenna array that has antenna spacing of 6.25cm (half­wavelength of 2.4GHz). This specific setting is expected to be affected by both the rich multipath in the indoor environment and the LOS component.

Latency vs. throughput in massive MIMO networks

We consider a multi­user massive MIMO system and minimize the average packet latency (from arrival to successfully delivery) with re­transmission. We leverage the massive MIMO spatial channel property to derive a computationally simple, closed­ form solution to minimize the latency. The computationally simple asymptotic is labeled as Large­-Array Simple Control (LASC), which captures the optimal transmission rate and error rate. The optimal transmission rate follows a simple thresholding rule for constant bit­rate traffic: The optimal transmission rate is 2A, where A is the packet arrival rate. If there are less than 2A packets in the queue, all waiting packets will be transmitted. We further show that error rate of the physical layer should adapt as , where is the CDF of the averaged channel across all M antennas; can be interpreted as the gap between link capacity and the traffic arrival rate. Interestingly, we find that latency­ optimality requires making the physical layer increasingly more robust as the base­station array size increases, i.e., to effectively reduce retransmission attempts.

Reusing cyclic-prefix (CP) intervals

We label the link that reuses the CP-intervals as CPLink as they use only the CP-intervals of the ongoing OFDM link labeled MainLink. We have designed and studied the zero­-knowledge CPLink, which drives CPLink interference at the MainLink receivers to the noise floor without any knowledge of MainLink receiver locations. First, we analytically guarantee that the zero­-knowledge capacity is non­zero, and design a scheme to choose the optimal transmit power and on­time (the duration for which CPLink remains on). Then, we numerically evaluate the zero­-knowledge CPLink. We study two types of CPLink. The first type is the full­ duplex CPLink, where the CPLink receiver is a full­duplex base­station. The second type, which is an additional use of CPLink, is the half­duplex CPLink, where any node other than the base­station is the CPLink receiver. Half­-duplex CPLink is a link between two nodes within the cell, similar to a device­to­device link. We have numerically evaluated the utility of CPLink in a setting similar to that of an LTE cellular network, with 20 MHz bandwidth. When the CP duration is about 7% of that of data symbol duration, the zero­-knowledge, full­-duplex CPLink can support data rates of up to 50 Mbps for nearby CPLink transmitters and up to ∼20 Mbps for the cell­edge CPLink transmitters, which are 2km away from the base­station.

Random access in massive MIMO networks

We show that random access delay can be reduced with the large number of spatial degrees of freedom. The key reason for the delay in random access is due to the collision event, where two users send packet overlapped in time and with the same packet format (e.g. modulation), which makes it impossible for the receiver to distinguish the two users apart. To reduce collision probability, LTE random access protocol allows users to randomly pick a spreading code (out of 64), thereby adding code diversity. Our results suggest that with large arrays, the spatial channel “code” of each user are also potentially separable, providing another avenue for the receiver to tell the two overlapping users apart in Angle­of­Arrival(AoA) space. However, exact separation is not possible due to multi­path and finite array resolution. Thus, the spatial channel codes of two users can also “collide.” We characterize the impact of array size and angular spread due to multipath on the collision probability of the system. We derive collision probability bounds that show that practical arrays sizes can provide 1.7× to 13× reduction in collision probability, depending on the scattering environment.


IN YEAR FOUR:

Experimental evaluation of antenna-domain massive MIMO channels

The key activity over the past year was experimental evaluation of channel correlations existing in antenna arrays and between users, and the resulting signal leakage up to 96 BS antennas. We con- ducted extensive channel measurements employing a 64-antenna base-station with two 2.4 GHz ISM bands, separated by 72 MHz, for both indoor and outdoor environment, including 8 indoor line-of-sight (LOS), 16 indoor non-line-of-sight (NLOS), 4 outdoor LOS and 21 outdoor NLOS mobile node locations. And in each mobile node location, we measured CSI for up to 5000 frames in outdoor environment and 250 frames in indoor environment.

FDD massive MIMO

We focus on FDD Massive MIMO CSIT acquisition for downlink beamforming. Recognizing that full downlink training, where each transmit antenna sends a pilot in time-orthogonal fashion, scales as O (M) for an M-antenna system, we investigate if uplink channel estimates provide any relevant information about downlink channel state information. Towards that end, we explore the use of angle-based channel modeling, where the overall channel response is modeled as a sum of scattered paths from different angles. We derive inspiration from past work that investigated reciprocity in FDD and discovered possible high angle-of-arrival (AoA) correlation over different bands but not conclusive, since the correlations of power angle spectrum estimated by a few antennas in one-dimension only are evaluated.

Understanding and alleviating the impact of pilot contamination

To obtain CSI in a scalable manner, time division duplex (TDD) massive MIMO systems use uplink pilots transmitted by UEs to estimate the uplink CSI and downlink CSI using channel reciprocity. However, this uplink pilot channel estimation approach is subject to pilot contamination, where adjacent cell UEs transmit non-orthogonal signals that interfere with the ongoing CSI estimation, and is predicted as a main limiting factor for massive MIMO system performance. The key activity over the past year was to study the impact of pilot contamination and propose detection method with zero startup cost and zero additional network overhead when pilot contamination happens, and validating with real channel measurements of 72-antenna BS.

Integrating full-duplex and massive MIMO

We has worked on developing an optimal MAC for an exciting new development called FlexRadio. This recent breakthrough in wireless MultiRF technology has introduced a new way to unify MIMO and full-duplex into a single framework with a fully flexible design. FlexRadio allows a wireless node to use an arbitrary number of RF chains to support transmission and reception, which makes MIMO and full-duplex subset configurations of FlexRadio. This new architecture has greatly changed the feasibility constraints in wireless networks, which makes the design of high performance MAC layer algorithms even more challenging. We are currently developing an architecture to integrate sub-6 GHz and mmWave technologies, where we incorporate the sub-6 GHz interface as a fallback data transfer mechanism to combat blockage and intermittent connectivity of the mmWave communications.


IN YEAR FIVE:

1. Multi-Hop Networking

In recent work we extend the we minimize the age of a single information flow in interference-free multihop networks. When preemption is allowed and the packet transmission times are exponentially distributed, we prove that a preemptive Last-Generated, First-Served (LGFS) policy results in smaller age processes across all nodes in the network than any other causal policy (in a stochastic ordering sense). In addition, for the class of New-Better-than-Used (NBU) distributions, we show that the non-preemptive LGFS policy is within a constant age gap from the optimum average age. In contrast, our numerical result shows that the preemptive LGFS policy can be very far from the optimum for some NBU transmission time distributions. Finally, when preemption is prohibited and the packet transmission times are arbitrarily distributed, the nonpreemptive LGFS policy is shown to minimize the age processes across all nodes in the network among all work-conserving policies (again in a stochastic ordering sense). Interestingly, these results hold under quite general conditions, including (i) arbitrary packet generation and arrival times, and (ii) for minimizing both the age processes in stochastic ordering and any non-decreasing functional of the age processes.

2. Integrating full-duplex and massive MIMO

At OSU, we have worked on developing an optimal MAC for an exciting new development called FlexRadio. This recent breakthrough in wireless MultiRF technology has introduced a new way to unify MIMO and full- duplex into a single framework with a fully flexible design. FlexRadio allows a wireless node to use an arbitrary number of RF chains to support transmission and reception, which makes MIMO and full-duplex subset configurations of FlexRadio. This new architecture has greatly changed the feasibility constraints in wireless networks, which makes the design of high performance MAC layer algorithms even more challenging. We are currently developing an architecture to integrate sub- 6 GHz and mmWave technologies, where we incorporate the sub-6 GHz interface as a fallback data transfer mechanism to combat blockage and intermittent connectivity of the mmWave communications. This work has been published in IEEE INFOCOM.

3. Integration of sub-6 GHz and mmWave technologies

At OSU, we are currently developing a multi-antenna architecture to integrate sub-6 GHz and mmWave technologies, where we incorporate the sub-6 GHz interface as a fallback data transfer mechanism to combat blockage and intermittent connectivity of the mmWave communications. We have also worked closely with PI Sabharwal at Rice to develop good beamforming and steering schemes and have published two papers in this area.

The following activities are being conducted extensively at both Rice and OSU.

4. FDD massive MIMO

We focus on frequency-division duplex (FDD) Massive MIMO Channel State Information at Transmitter (CSIT) acquisition for downlink beamforming. Recognizing that full downlink training, where each transmit antenna sends a pilot in time-orthogonal fashion, scales as O(M) for an M-antenna system, we investigate if uplink channel estimates provide any relevant information about downlink channel state information. Towards that end, we explore the use of angle-based channel modeling, where the overall channel response is modeled as a sum of scattered paths from different angles. We derive inspiration from past work that investigated reciprocity in FDD and discovered possible high angle-of-arrival (AoA) correlation over different bands but not conclusive, since the correlations of power angle spectrum estimated by a few antennas in one-dimension only are evaluated.

5. Minimizing Delay in Massive MIMO Communications

One fundamental challenge in 5G URLLC is how to optimize massive MIMO systems for achieving low latency and high reliability. A natural design choice to maximize reliability and minimize retransmission is to select the lowest allowed target error rate. However, the overall latency is the sum of queueing latency and retransmission latency, hence choosing the lowest target error rate does not always minimize the overall latency. In this work, we minimize the overall latency by jointly designing the target error rate and transmission rate adaptation, which leads to a fundamental tradeoff point between queueing and retransmission latency.

Broader Impacts: This work had resulted in a number of broader impacts. (i) Provided opportunities for a female graduate student to conduct research in an exciting area; (ii) Allowed for research results to be integrated in a hybrid (undergrad-grad) class on wireless networks; (iii) Facilitated interactions with industry and other universities.


Broader Impacts

    Educational Activities:

    Results from this research were integrated into a wireless networking course. This course typically has a mix of undergraduate and graduate students and is a project heavy course. Group projects were proposed that allowed students to leverage the latest wireless networking designs into their solutions.

    Outreach:

    One female graduate student currently works on problems related to this project and has developed wireless networking solutions for multi-antenna and mmWave based systems. Our PhD students have also done summer internships at various companies and after graduation have joined major wireless telecommunications companies (e.g., AT&T, Qualcomm, etc.).


Publications

  • A. M. Bedewy, Y. Sun, and N. B. Shroff, “The Age of Information in Multihop Networks” IEEE/ACM Transactions on Networking Volume 27, Issue 3, June 2019 pp 1248–1257
  • A. M. Bedewy, Y. Sun, R. Singh and N. Shroff, "Low-Power Status Updates via Sleep-Wake Scheduling" in IEEE/ACM Transactions on Networking, vol. 29, no. 05, pp. 2129-2141, 2021
  • J. Pan, A. M. Bedewy, Y. Sun, and N. B. Shroff, “Minimizing Age of Information via Scheduling over Heterogeneous Channels,” ACM MobiHoc’21, Shanghai, China, July 2021
  • G. Yao, A. M. Bedewy, and N. B. Shroff, “Battle between Rate and Error in Minimizing Age of Information,” ACM MobiHoc’21, Shanghai, China, July 2021
  • Xu Du, Yin Sun, Ness Shroff, and Ashutosh Sabharwal (2020). Balance Queueing and Retransmission: Latency-Optimal Massive MIMO Design. IEEE Transactions on Wireless Communications. Status = ACCEPTED; Acknowledgment of Federal Support = Yes ; Peer Reviewed = Yes.
  • G. Yao, M. Hashemi, and N. B. Shroff (2019). Integrating Sub-6 GHz and Millimeter Wave to Combat Blockage: Delay-Optimal Scheduling. IEEE WiOpt 2019. Status = PUBLISHED; Acknowledgment of Federal Support = Yes ; Peer Reviewed = Yes.
  • J. Kwak and N. B. Shroff (2019). Simulated Annealing for Optimal Resource Allocation in Wireless Networks with Imperfect Communications. IEEE Allerton.
  • Jian Ding and Ranveer Chandra (2019). Towards Low Cost Soil Sensing Using Wi-Fi. ACM MobiCom.
  • Z. Qian, Y. Yang, K. Srinivasan, and N. B. Shroff (2019). Joint Antenna Allocation and Link Scheduling in FlexRadio Networks. IEEE INFOCOM. Status = PUBLISHED; Acknowledgment of Federal Support = Yes ; Peer Reviewed = Yes.
  • C.-Y. Yeh. Feasibility of passive eavesdropping in massive mimo: An experimental approach. (2018). Rice University Master Thesis.
  • Xing Zhang, Lin Zhong, and Ashutosh Sabharwal (2018). Directional Training for FDD Massive MIMO. IEEE Transactions on Wireless Communication.
  • C.­Y Yeh and E. Knightly (2018). Feasibility of Passive Eavesdropping in Massive MIMO: An Experimental Approach. in Proceedings of IEEE CNS.
  • F. Ahsan and A. Sabharwal (2018). Leveraging massive MIMO spatial degrees of freedom to reduce random access delay. 51st Asilomar Conference on Signals, Systems, and Computers.
  • Clayton Shepard, Rahman Doost­Mohammady, Ryan E. Guerra, and Lin Zhong (2017). Demo: ArgosV3: an efficient many­antenna platform. ACM MobiCom.
  • J. Liu, A. Eryilmaz,, N. B. Shroff, and E. S. Bentley (2017). Understanding the Impacts of Limited Channel State Information on Massive MIMO Cellular Network Optimization. IEEE Journal on Selected Areas in Communications (JSAC).
  • N. M. Gowda and A. Sabharwal, (2018). JointNull: Combining Partial Analog Cancellation with Transmit Beamforming for Large­Antenna Full­Duplex Wireless Systems. IEEE Transactions on Wireless Communications. 17 (3).
  • S. Wang and N. B. Shroff (2018). Towards Fast­Convergence, Low­Delay and Low­Complexity Network Optimization. ACM Sigmetrics.
  • Xing Zhang, Lin Zhong, and Ashutosh Sabharwal. Directional Training for FDD Massive MIMO. to appear in IEEE Transactions on Wireless Communication
  • Ahmed M. Bedewy, Yin Sun, and Ness B. Shroff (2017). Optimizing Data Freshness, Throughput, and Delay in Multi-Server Information-Update Systems. IEEE ISIT.
  • Clayton Shepard, Abeer Javed, Ryan Guerra, Jian Ding, and Lin Zhong (2016). Many-Antenna MU-MIMO Channel Measurements. IEEE ASILOMAR Conference.
  • F. Wu, Y. Yang, O. Zhang, K. Srinivasan, and N. B. Shroff (2016). Anonymous-Query based Rate Control for Wireless Multicast: Approaching Optimality with Constant Feedback. ACM MobiHoc.
  • Fatima Ahsan and Ashutosh Sabharwal, (2017). Leveraging Massive MIMO Spatial Degrees of Freedom to Reduce Random Access Delay. under review for Asilomar Conference on Signals, Systems and Computers.
  • J. Liu, A. Eryilmaz, N. B. Shroff, and E. Bentley (2016). “Understanding the Impact of Limited Channel State Information on Massive MIMO Network Performances. ACM MobiHoc.
  • J. Liu, A. Eryilmaz,, N. B. Shroff, and E. S. Bentley (2017). Understanding the Impacts of Limited Channel State Information on Massive MIMO Cellular Network Optimization. IEEE Journal on Selected Areas in Communications (JSAC).
  • P. Nayak, M. Garetto, and E. Knightly (2017). Multi-User Downlink with Single-User Uplink can Starve TCP. IEEE INFOCOM.
  • A. Flores, S. Quadri, and E. Knightly (2016). A Scalable Multi-User Uplink for Wi-Fi. Proceedings of NSDI
  • A. Kwong and A. Sabharwal (2015). Overcoming Conjugate Beamforming Limitations with Side-Channel Cooperative Decoders. IEEE Asilomar Conference on Signals, Systems and Computers (ASILOMAR).
  • Clayton Shepard, Abeer Javed, and Lin Zhong (2015). Control channel design for many-antenna MU-MIMO. Proc. ACM Int. Conf. Mobile Computing and Networking (MobiCom).
  • Clayton Shepard, Abeer Javed, Ryan Guerra, Jian Ding, and Lin Zhong (2016). Many-Antenna MU-MIMO Channel Measurements. IEEE ASILOMAR Conference.
  • J. Liu, A. Eryilmaz, N. B. Shroff, and E. Bentley (2016). Understanding the Impact of Limited Channel State Information on Massive MIMO Network Performances. Proc. ACM MobiHoc
  • Xing Zhang, John Tadrous, Evan Everett, Feng Xue, and Ashutosh Sabharwal (2015). Angle-of-arrival based beamforming for FDD massive MIMO. IEEE Asilomar Conference on Signals, Systems and Computers (ASILOMAR).
  • F. Wu, Y. Yang, O. Zhang, K. Srinivasan, and N. B. Shroff (2016). Anonymous-Query based Rate Control for Wireless Multicast: Approaching Optimality with Constant Feedback. ACM MobiHoc.
  • Y. Sun, E. Uysal-Biyikoglu, R. Yates, C. E. Koksal, and N. B. Shroff (2016). Update or Wait: How to Keep Your Data Fresh. IEEE Infocom.
  • Z. Qian, B. Ji, K. Srinivasan, and N. B. Shroff (2016). Achieving Delay Rate-function Optimality in OFDM Downlink with Time-correlated Channels. IEEE Infocom.