Mesh-based Sensor Network Backbones
We have designed Sprinkler, a reliable data dissemination service for wireless embedded devices which are
constrained in energy, processing speed, and memory. Sprinkler embeds a virtual grid over the network whereby
it can locally compute a connected dominating set of the devices to avoid redundant transmissions and a
transmission schedule to avoid collisions. Sprinkler transmits O(1) times the optimum number of packets
in O(1) of the optimum latency; its time complexity is O(1). Sprinkler is tolerant to fail-stop and state
corruption faults. Thus, Sprinkler is suitable for resource-constrained wireless embedded devices. We evaluate
the performance of Sprinkler in terms of the number of packet transmissions and the latency, both in an outdoor
and indoor environment. Based on outdoor and indoor testbed experiments, we show that Sprinkler is not only
energy efficient as compared to existing schemes, but also has less latency. Further, the energy consumption of
nodes and the latency grows linearly as a function of newly added nodes as the network grows larger.
In the context of IEEE 802.11b network testbeds, we have examined the differences between unicast and
broadcast link properties, and shown that the inherent difficulties in precisely estimating unicast link properties
via those of broadcast beacons even if we make the length and transmission rate of beacons be the same as
those of data packets. To circumvent the difficulties in link estimation, we proposed to estimate unicast link
properties directly via data traffic itself without using periodic beacons. To this end, we designed a data-driven
routing protocol Learn on the Fly (LOF). LOF chooses routes based on ETX/ETT-type metrics, but the
metrics are estimated via MAC feedback for unicast data transmission instead of broadcast beacons. Using a
realistic sensor network traffic trace and an 802.11b testbed of 195 Stargates, we have experimentally compared
the performance of LOF with that of beacon-based protocols, represented by the geography-unaware ETX and
the geography-based PRD. We found that LOF reduces end-to-end MAC latency, enhances energy efficiency,
and improves network throughput, which demonstrate the feasibility and the potential benefits of data- driven
link estimation and routing.
Most Relevant Papers