A Platform for WEbS (wireless embedded sensor/actuator) systems David Culler Eric Brewer Dave Wagner
Enable NEST Research (Network Embedded Software Technology) Goal: enable research in algorithms, synchronization, real-time systems Low-cost, large-scale experimentation Extensible Tiny OS (byte code) Infrastructure services Simulation environment Development Environment
+ Research Itself Power management Networking Security Resilient Aggregation –collect data with noise, failures and adversaries FSM programming & composition Macrocomputing: programming a collection
Platforms Current Phase 1: 6 months => algorithm studies –Mote++, MEMS sensors Phase 2: 20 months => composition –ARM, Bluetooth physical –integrated system Workshops for both phases
Nodal Communication Local multicast event-driven reception intelligent pruning aggregation buffer management remote programming/debugging/upgrade key management synchronized logging (trace extraction)
Adversarial Simulation Large-scale mote simulator Detecting “composition” bugs Target failure: search for bugs –test race conditions automatically –pick orders that consume resources –more efficient than random-walk testing –simulator is an adversary… Hybrid simulator/testbed
Development Environment Make FSMs work –event-driven programming –composition –debugging –visualization Same four issues for collections of WEbS!
Macrocomputing How to program a large collection? –Single program, multiple data but errors and probabilistic behavior –“global” variables that reflect collections need to handle error propagation –scatter/gather for collections? –online query processing? Need multi-WEbS abstractions
Summary We have rare advantages… –MEMS center –Millennium cluster (services & simulation) –working hardware –event-driven programming experience –early wins in algorithms and security –language expertise –dense distributed systems expertise This should be fun