Dynamic Spectrum Access under Uncertainty: Theory, Algorithm Development, and Evaluation

NeTS Small

List of personnel

  1. Principal Investigator: Ness Shroff

  2. Co-Principal Investigator: Eylem Ekici

  3. Graduate Students: Ahmed Bedewy, Sherif ElAzzouni, Fang Liu, Fei Wu, and Xingyu Zhou

Project goals

The major goal of this project is to understand at a fundamental level how to efficiently manage dynamic spectrum access in the presence of uncertainty of spectral availability, channel states, and secondary traffic. Our approach is to be able to control this uncertainty appropriately via collaborative sensing and information sharing, taking into account the gains and costs in a system composed of Primary and Secondary Providers.

Major activities

    In the latest reporting period, we have studied the following problems:

  1. Truthful Mobile Crowd-sensing for Strategic Users with Private Qualities

    We study the problem of mobile crowd-sensing; a key enabling technology for large scale spectrum sensing in dynamic spectrum access problems. The idea of mobile crowd-sensing is to use a large number of mobile users to measure a quantity of interest, and leverage data aggregation to obtain accurate estimates. The requester of the information has several challenges to solve for the crowd-sensing task to be beneficial. The most essential being: How to assign tasks to different mobile users, in particular, which mobile users should the requester choose to perform the task; On one hand, using a large number of users may increase the accuracy of the sensing, on the other, the requester has to compensate those users for sensing costs, as well as incentivizing them to sense. One problem of assigning tasks to users based on their quality of information is that the quality may depend on private information such as location. Another problem is that users may decide to manipulate the quality of their information or the effort they’ve put in the task in terms of time, energy, resources, ...etc. While other works in the literature have focused on either truthful crowd sensing with private costs, mechanism design for hidden actions or quality-aware crowd-sensing, we have focused on the on the setting where quality is a user’s private information that is unknown to the requester.

  2. Delay-Optimal Buffer-Aware Scheduling With Adaptive Transmission

    We aim to obtain the optimal tradeoff between the average delay and the average power consumption in a spread spectrum sensing and communication system. The power-delay tradeoff is well known in delay-sensitive communication system. In particular, the transmitter can wait for favorable channel conditions to transmit at high rate with low power, or the transmitter can transmit as soon as there is data in its queue, which minimizes the delay at the cost of high power consumption. In our system, the arrivals occur at each time slot according to a Bernoulli arrival process, and are buffered at the transmitter waiting to be scheduled. We consider a finite buffer and allow the scheduling decision to depend on the buffer occupancy. In order to capture the realism in communication systems, the transmission power is assumed to be an increasing and convex function of the number of packets transmitted in each time-slot.

  3. Pricing for Past Channel State Information in Multichannel Cognitive Radio Networks

    We consider a Dynamic Spectrum Access (DSA) problem where the Secondary User (SU) can use both spectrum sensing and database access to make decisions on accessing primary channels. Existing works on the literature have usually considered the sensing problem using either database access only or spectrum sensing techniques such as energy detection only. However, we see those two approaches as complementary to each other as database information is more reflective of long term statistics, channel history, channel trends…etc., whereas spectrum sensing attempts to measure immediate availability. We focus on the case where Primary Providers (PP) can sell their channel histories to the SU to aid the SU in making channel access decisions. We investigate the interplay between the PPs and the SUs through their pricing and buying decisions for this information, in the presence of sensing inaccuracy, i.e., false alarm and miss detection. As a first step, we consider a CR network with one SU and two PPs, where each PP uses a separate spectrum band and then extend the results to multiple SUs. To incite the PPs to participate in and voluntarily provide their activity information, each PP with an assigned spectrum band can set the price of its information about its previous activities. For each channel, the SU can buy the information, and use it to figure out the most-likely vacant channel. We model the channel information market as a two- stage sequential game (i.e., Stackelberg game), and we investigate the pricing behavior of the PPs for their channel activity information, and the SU decisions in the single SU case. We first consider the case when PPs are competitive, find the Nash equilibrium, then we consider the case when the PPs can cooperate and show that the cooperation may decrease social welfare (though it improves the revenues of the PPs). We extend the above results to multiple SUs, and investigate the impact of multiple SUs. We verify our results through numerical simulation with various parameters.

  4. Using Dynamic Spectrum for Delay-Aware Predictive Offloading

    We consider dynamic spectrum as a means to aid cellular networks; In particular we study the case where mobile users can offload some of their data to dynamic unassigned spectrum. It is a well-known fact that demand on higher data-rates is straining the cellular infrastructure. One proposed solution is offloading to Wifi networks, as the Wifi spectrum is “cheap” in the sense that providers do not meter Wifi usage. Similarly, dynamic spectrum holes can be leveraged to offload some data to ease congestion on cellular spectrum. The differences between cellular and dynamic spectrum is that cellular spectrum is ubiquitous, reliable in terms of rate and valuable whereas dynamic spectrum is intermittent, unreliable and cheap to use. While many papers have studied different modes of offloading, delayed-offloading remains the most attractive option, whereby the mobile user can delay transmission of some delay-tolerant applications until cheap spectrum appears. Those studies however lack theoretical efforts to understand and optimize performance. We consider a mobile user transmitting several elastic flows with different delay sensitivities. We assume that up to a small window in the future, the mobile transmitter can accurately predict the dynamic spectrum availability. We formulate the problem as a Discounted Rate Utility Maximization (DRUM) problem, where delay sensitivity is modeled as a discount on the utility function, the optimization variables are the cellular and the dynamic rates every time slot. Cellular usage is penalized linearly according to a cost set by the provider to discourage extensive cellular usage; this corresponds to cases of “metered” cellular usage. The intermittent dynamic spectrum capacity appears as a random variable each time slot drawn from a possibly time dependent distribution. We propose a low complexity online solution to the problem with the aid of prediction, prove that under some mild conditions, the online algorithm achieves a constant-competitive ratio compared to the prescient offline solutions. We use simulations to show that the proposed solution gives the expected behavior where delay tolerant flows transmit in high rate bursts when dynamic spectrum is available whereas delay sensitive flows maintain a consistent rate irrespective of dynamic spectrum capacity.


  1. J. Liu, A. Eryilmaz, N. B. Shroff, and E. S. Bentley, “Understanding the Impacts of Limited Channel State Information on Massive MIMO Cellular Network Optimization”, IEEE Journal on Selected Areas in Communications (JSAC), accepted for publication.

  2. S. Kang, C. Joo, J. Lee, and N. B. Shroff, “Pricing for Past Channel State Information in Multichannel Cognitive Radio Networks”, IEEE Transactions on Mobile Computing, accepted for publication.

  3. X. Chen, W. Chen, J. Lee, and N. B. Shroff, “Delay-Optimal Buffer-Aware Scheduling with Adaptive Transmission,” IEEE Trans. on Communications (TCOM), accepted for publication.

  4. J. Lee and E. Ekici, “Sensor Selection Under Correlated Shadowing in Cognitive Radio Networks”, IEEE COMMUNICATIONS LETTERS, Vol. 21, 2017.

  5. X. Gong, and N. B. Shroff, “Truthful Mobile Crowd-sensing for Strategic Users with Private Qualities”, IEEE 15th Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks (WiOpt'17), Paris, France, May 2017.

  6. C. Joo, X. Lin, J. Ryu, and N. B. Shroff, “Distributed Greedy Approximation to Maximum Weighted Independent Set for Scheduling with Fading Channels”, IEEE/ACM Trans. on Networking (ToN), vol. 24, no. 3, pp. 1476-1488, Jun. 2016.

  7. S. Kwon, Y. Kim, and N. B. Shroff, “Analysis of Connectivity and Capacity in One-Dimensional Vehicle-to-Vehicle Networks”, IEEE Trans. on Wireless Communications (TWC), Vol. 15, Dec. 2016.

  8. S. Li, E. Ekici and N. B. Shroff, “Throughput-Optimal Queue Length Based CSMA/CA Algorithm for Cognitive Radio Networks”, IEEE Trans. on Mobile Computing, Vol 14, no. 5, May 2015, pp. 1098-1108.

  9. S. Li, Z. Zheng, E. Ekici, and N. B. Shroff, “Maximizing System Throughput by Cooperative Sensing in Cognitive Radio Networks”, IEEE/ACM Trans. on Networking, vol. 22, no. 4, Aug. 2014, pp. 1245-1256.

  10. W. Ouyang, A. Eryilmaz, and N. B. Shroff, “Downlink Scheduling over Markovian Fading Channels”, IEEE/ACM Trans. on Networking (ToN), vol. 15, Issue 4, Mar. 2016, pp. 909 - 923.

  11. Y. Kim, K. Lee, and N. B. Shroff, “On Stochastic Confidence of Information Spread in Opportunistic Networks”, IEEE Trans. on Mobile Computing, vol. 15, no. 4, pp. 909-923, Apr. 2016.

  12. Z. Qian, B. Ji, K. Srinivasan, and N. B. Shroff, “Achieving Delay Rate-function Optimality in OFDM Downlink with Time-correlated Channels”, IEEE INFOCOM'16, San Francisco, CA, Apr. 2016.

Broader Impacts:

The outcomes of this project have been presented at conferences, major universities, and to major industries like Qualcomm and Samsung.

Ohio State Engineering:
Excellence • Impact • Innovation