There is an increasing need for real-time wireless communications in medical care, surveillance and vehicle-to-vehicle networks. Currently, such embedded wireless networks suffer from short battery life, unpredictable real-time performance and poor scalability. To enable multi-hop wireless network applications with predictable and well-defined real-time properties, I present a cross-layer approach to design networked wireless systems. My research is applied across three segments spanning broadband wireless transceivers (MEERA), low-cost wireless sensor networks (FireFly) and highly mobile vehicular networks (GrooveNet).

The unifying theme in my research is that all systems are designed with well-defined real-time properties and include tightly coupled time synchronization. Here's a brief overview of the components of my recent research. You can find more details in each section.

MEERA is a Methodology for Energy-Efficient Resource Allocation for broadband wireless transceivers. MEERA determines, at run-time, the optimal settings across RF electronics, communication tradeoffs and the link layer to deliver delay-sensitive data streams over a time-varying channel. MEERA increases the system lifetime by a factor of 3-to-8.

Click here for an overview of MEERA

See related publications

FireFly is a dual-radio sensor networking platform with predictable and near-optimal node lifetimes (up to 2 years on two AA batteries). We achieve this through tight hardware-based global time-synchronization with sub-100us synchronization accuracy. A 42-node network was deployed in a Coal Mine for miner rescue and multi-hop voice streaming.
Click here for an overview of Firefly and a cool video of HW-based global time sync with an AM radio [8MB]

Or go to the FireFly Website

GrooveNet: Vehicular wireless networks will make driving safer, more efficient and more enjoyable.  I have built GrooveNet, a vehicular network virtualization platform where the same network models are used both in simulation and in real mobile test-beds, while allowing interaction between real and virtual vehicles. By accurately modeling inter-vehicular communication among thousands of vehicles within a street map-based topography, GrooveNet facilitates protocol analysis, rapid in-vehicle deployment and model validation. GrooveNet is open-source and is being used by over 18 research institutions for protocol design in vehicular networks. Click here to go to the GrooveNet Website

MAX is a time-division-multiplexed resource allocation framework for multi-hop networks with regular topologies. MAX tiling delivers optimal end-to-end throughput across arbitrarily large regularly structured networks while providing bounded delay. It outperforms CSMA-based random access protocols by a factor of 5 to 8. The MAX approach also supports network services including flexible uplink and downlink bandwidth management, deterministic route admission control, and optimal gateway placement. MAX has been implemented on IEEE 802.15.3 embedded nodes and a test-bed of 16 nodes has been deployed both indoors and outdoors.

  Click here to see a demonstration of Distributed Tile Replication (Requires Windows)

  Click here to see a demonstration of Routing with SuperNodes on a Grid Network

Wireless Jamming Avoidance and 2-Factor Authentication

Electromagnetic jamming presents a serious security attack in sensor networks as it is easy and efficient to perform on existing sensor network systems and protocols. In CSMA-based protocols, jamming attacks result in denial of service by preventing transmitters from sending packets and also drain the energy of receivers. In TDMA-based networks with fixed and periodic packet exchanges, it is possible to completely jam a receiver while using an energy-efficient pulse jamming attack. The focus of this project was to devise a jamming avoidance scheme and implement it using a slot schedule randomization scheme to obfuscate any patterns of channel activities from a jammer. Our jamming avoidance scheme employed a secure 1-way SHA1 key generation at the gateway that was periodically and securely distributed (every 5 seconds) to all nodes. Each node concatenated the key with its node id and computed a 160 bit random number sequence using HMAC-SHA1 to determine the 4ms transmit slots and precedence for each TDMA frame. In addition each node, computed the schedules of all its neighbors and resolved its transmit and receive schedule based on slot precedence. Jamming avoidance was implemented on the Atmel ATMEGA32L based CMU FireFly sensor nodes. We demonstrated the performance using a real test-bed. Our scheme reduces the probability that a particular node's packet is jammed to 0.002 and incurs only a 2-byte overhead in each gateway broadcast every 5 seconds.

  Experimental video [30MB Quicktime] of Randomized, but coordinated, slot scheduling. The top two signal is the transmitter and the middle signal is the receiver. The bottom signal, in pink, is just to show you the reference of the time slots and TDMA frames.