Lara B. Deek

Multi-User MIMO (MU-MIMO) in IEEE 802.11ac WLANs

The emerging IEEE 802.11ac standard provides another leap in supported 802.11 data rates, with up to 600Mbps data rates compared to 300Mbps for 802.11n and 54Mbps for legacy 802.11a/b/g clients. IEEE 802.11ac adds new features over 802.11n, namely wider channel widths of up to 160MHz, MU-MIMO transmission schemes, and a higher modulation scheme (256QAM). Of the most interesting additions is MU-MIMO. Our work attempts to propose a standard-compliant wireless system that exploits the benefits and features of MU-MIMO, building on fundamental communication theory concepts and techniques. We implement and evaluate our system primarily on GNU Radio and the USRP.

The IEEE 802.11n Standard in WLANs

IEEE 802.11n revolutionized WiFi technologies by adding major upgrades to legacy IEEE 802.11a/b/g clients. These upgrades feature MIMO transmission schemes and channel bonding two 20MHz channels into a single 40MHz channel. These features provide greater opportunities to exploit the existing bandwidth. Although each feature alone provides benefits over previous legacy clients, it is only through understanding the behavior of these features can efficient strategies of utilizing them be proposed. In our work, we build an understanding of 802.11n features in real testbed environments and design wireless systems that exploit their benefits, using off-the-shelf wireless cards.

Note: The open-source community has shown interest and engaged us in active discussion on our 802.11n solutions, and particularly our joint rate and channel width adaptation solution, dubbed ARAMIS [pdf]. I anticipate contributing my solution to the open-source community!

Online Spectrum Auctions for White Spaces

Spectrum auctions have emerged as a popular and effective model to allocate white space spectrum. In wireless environments, users can acquire and release bandwidth at any time and in real time. We therefore propose an online spectrum auction model, which provides users with the flexibility to obtain spectrum in real-time. Such flexibility, however, creates vulnerabilities to bidder manipulations, whereby bidders can rig their bid to gain an unfair advantage. In our work, we propose an online spectrum auction Topaz which distributes spectrum efficiently while discouraging bidders from misreporting their bids or availability times. Furthermore, Topaz introduces a preemptive mechanism which improves auctioneer revenue at a minimum cost to spectrum efficiency.

Locality of Interest in Online Social Networks

Online social networks have gained immense popularity and a large user-base worldwide. In this work, we analyze the behavior of Facebook users through real-world OSN trace and show that a significant percentage of user interactions are local. These findings motivate the need for a distributed OSN architecture that improves performance for geographically dispersed users.

Concurrent Use of Heterogeneous Networks

Modern mobile devices are equipped with multiple wireless technologies. With this development, it is not a surprise to find a single user with many connection opportunities. Traditionally, devices have either used one interface at a time or have primarily used one interface for particular transmissions, such as either data packets or control packets. In our work, we focus on methods of concurrently using multiple radios, such as WiFi, cellular, and satellite, while addressing the major challenges of doing so. The challenges include (1) recognizing and monitoring the characteristics of each wireless technology, such as performance and state, (2) determining efficient assignments of transmissions to radio interfaces, (3) identifying the utilization cost factor, and (4) determining the way by which networks can be utilized to better serve different network applications.

Application-Aware Wireless Systems for Mobile Environments

Applications running on mobile devices, such as web, or email, suffer greatly from degraded performance due to intermittent connectivity. This phenomenon is exemplified by the fact that an incompletely transmitted data transfer cannot be viewed by the receiver due to missing packets. Therefore, any incomplete transmission is made inaccessible and useless. We therefore design an adaptive system which predicts and adapts packet size to changing network conditions, namely connectivity and throughput. The performance of the system was measured in terms of the amount of viewable data the user receives per connection period.