BBN Technologies

The field of mobile multi-hop, or ad hoc, networking has been evolving since the late 1960s, though we are just now seeing mobile ad hoc networking solutions readily available from a number of places. BBN has been at the forefront of designing, developing, and building mobile ad hoc networks since the very earliest days.

Ad hoc networking allows communications over multiple nodes without a fixed infrastructure and allows a distributed, autoconfiguring routing architecture that can adapt to any change. These capabilities are most valuable in situations where energy is a constrained resource. Whether a sensor system, a individual soldier radio, or a mobile robot, reducing the energy requirements for an ad hoc network is critical.

Most "energy efficient" ad hoc networks are designed piece-meal — an improved routing protocol here, a lower power radio there — all without thinking about how the individual parts need to work together to maximize their total system-wide gains. Furthermore, many traditional solutions for reducing the energy used by an ad hoc network are designed to work in only one small, particular niche environment.

BBN has designed an ad hoc network protocol stack that is unique in that it provides true cross-layer optimizations from the transport layer all the way down to the radio, providing the absolute best performance without the constraints of layering or pre-existing networking traditions. Furthermore, our design relies only on off-the shelf parts for its ground-breaking operation, though it can utilize new advances as they become available and cost-effective.

Our approach is radically different than what others have done in the past because we exploit a variety of factors at every point in the design:

• At the physical layer, we exploit the ability of low data rate radios to require less energy per bit, due to real-world receiver design issues, than their high data rate counterparts. Furthermore, we exploit time and frequency-coded modulation to allow multi-packet reception, nearly eliminating MAC control overhead.
• At the MAC layer, we have been exploiting new mathematical structures to provide uncorrelated, but completely deterministic wakeup sequences that require very little over-the-air traffic to express wakeup schedules among neighbors. This also allows us the flexibility to dynamically adapt our energy usage at the MAC layer to the exact traffic load.
• For routing, our system exploits some of the most thorough analytic recent work in routing to limit control traffic to the places where it is needed for carrying user traffic.
• At the End-to-End layer, we use cross-layer techniques such as caching and intertwined control loops to minimize end-to-end retransmissions, allow mid-path NAKs to flow back to sources when end-to-end links break, and allow the transport layer itself to reroute packets or change MAC packets dynamically in response to local congestion.

All of these innovations can operate independently over existing designs, but when used together they provide gains that are greater than the sum of their parts. Our designs have shown 50-300X reduction in the energy required per node without significantly affecting applications constraints such as delay. BBN is moving forward with making the next generation of low energy networking a reality.