Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


SwapnaDissertation.pdf (1.54 MB)

ETD Abstract Container

Abstract Header

Control of Large Scale Networked Systems Under Uncertainty

Buccapatnam Tirumala, Swapna
http://rave.ohiolink.edu/etdc/view?acc_num=osu1416764356

Abstract Details

2014, Doctor of Philosophy, Ohio State University, Electrical and Computer Engineering.
The rapid growth and popularity of smart mobile devices, wireless networks, and online social networks has transformed the nature of social interactions as well as created a new platform for decision-makers (sellers, policymakers, etc.) to interact with users. New services developed on these platforms require not only faster data rates but also lower communication delays. These new services can also be improved by personalizing them to the users' preferences and learning from their social interactions. This dissertation investigates control policies for large social and wireless networks. It aims to overcome challenges due to scale and uncertainty that plague these systems. In the case of social networks, we develop control policies that can be used by decision-makers to provide better personalized services to users while leveraging the externalities that arise due to the users' social interactions. Specifically, this dissertation addresses the issue of unknown user preferences as well as the users' lack of global network knowledge. For users with limited network information, we develop dynamic pricing policies for the seller that guarantee both significantly better revenue for the seller as well as improved utility for the buyers over existing static policies. We also develop policies that the seller can use to better target consumers when the user preferences are unknown to the seller. These policies leverage the user interactions to accelerate the inference of these unknown preferences. Motivated by these policies, we explore more general reward maximization problems under uncertainty where additional information can be gleaned from the network structure. For example, this additional information is due to user interactions in social networks while in sensor networks, it is due to aggregation of data from nearby sensors. For wireless networks, we investigate how to lower delay and maximize throughput in the presence of time-varying uncertainty of wireless communication channels. We investigate these performance metrics for two important communication techniques, namely, retransmissions and network coding. One of the most common ways to ensure reliable communication is to retransmit data packets that have not been acknowledged by the receiver. For the fundamental setting of a single source-destination pair connected by a time-varying channel, we reveal the surprising result that even when the maximum packet size is small, the variability in packet sizes can be amplified by retransmission-based protocols, leading to poor delay and throughput performance. This result implies that coding mechanisms such as random network coding should be used to prevent or minimize retransmissions. In fact, recently, network coding schemes have been proposed to support significantly higher data rates than traditional methods. In this dissertation, we identify a critical scaling of batch sizes used in random linear network coding to ensure improved delay performance as well, even in the presence of time-varying channels. This scaling ensures optimal throughput and delay performance simultaneously for large scale network broadcasts with time-varying wireless links.
Ness Shroff (Advisor)
Atilla Eryilmaz (Advisor)
Can Emre Koksal (Committee Member)
Mikhail Belkin (Committee Member)
206 p.
Computer Science; Electrical Engineering; Systems Design

Recommended Citations

Citations

  • Buccapatnam Tirumala, S. (2014). Control of Large Scale Networked Systems Under Uncertainty [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1416764356

    APA Style (7th edition)

  • Buccapatnam Tirumala, Swapna. Control of Large Scale Networked Systems Under Uncertainty. 2014. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1416764356.

    MLA Style (8th edition)

  • Buccapatnam Tirumala, Swapna. "Control of Large Scale Networked Systems Under Uncertainty." Doctoral dissertation, Ohio State University, 2014. http://rave.ohiolink.edu/etdc/view?acc_num=osu1416764356

    Chicago Manual of Style (17th edition)

osu1416764356
466
© 2014, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.