RDSP: Research Group - Resource constrained Distributed Stream Processing
We research an integrated approach for supporting in-network sensor data processing in dynamic and heterogeneous sensor networks.
The concept relies on data stream processing techniques that define and optimize the distribution of queries and their operators.
The Project is approached from three different directions:
- DSAM (Data Stream Application Manager):
DSAM is a central manager of data stream applications. Its main goal is the integration of heterogeneous Stream Processing Systems (SPSs).
As it is not practicable to send all data to a central site, DSAM supports distributed query processing (also called in-network query processing in WSNs).
DSAM achieves this integration by using the MDA approach.
We use abstract queries that describe the global queries and correspond to the PIM of the MDA.
- Operating Systems:
We are developing a framework for stateful migration of modules.
This framework allows us to migrate data stream operators from one node to another one without losing its state.
- Energy cost and lifetime estimation:
There are many reasons to migrate DS-Operators from one node to another. One of the most important in WSNs is the lifetime of the network.
To estimate a battery-driven mote's lifetime a sophisticated estimation of energy costs is needed. This must not only include the energy needed to
run a node, but also the costs of migrating operators. Based on these estimations the DSAM can decide to reorganize the DS-operators.
The RDSP research project is a cooperation of three different research groups:
MSP430 on-node linking and migration
As part of our research we developed an optimized linker "Minilink", the possibility to migrate
Protothreads, called Migmem, and some other tools. The sources can be found at the Contiki Projects Community.
Simulation of real deployments
Several tools for Cooja are under development. Upon them are RealSim (paper /
code) and
CoojaTrace (paper / code).
Links
