1. Distributed Estimation in Mobile Sensor Network
A sensor network is a collection of small sensors that are equipped with sensing, processing and communication (typically wireless) capability. A lot of applications of sensor networks have been implemented in variety of areas:
Time Synchronization:
Time synchronization in sensor network is extremely important in effective use of sensor network. The utility of data collected and transmitted by sensor nodes depends directly on the accuracy of the time-stamps. In TDMA based communication schemes, accurate time synchronization is required for the nodes to communicate with each other. Operation on a pre-scheduled sleep-wake cycle for energy conservation and lifetime maximization, and detection of multiple events also require accurate knowledge of a global time. However, clocks in nodes run imperfectly due to drifting of quartz clock. Each node only knows its local time, which is different from the global time.
For instance, in the figure above, two nodes can detect time instant of an object breaking into their detection range as shown in the circle. An object first passes through sensor 1 at global time 2:29pm, and then sensor 2 at 2:30pm. However, what the two sensors actually read are 2:32pm and 2:29pm according to their local clocks. Therefore, nodes will make a wrong conclusion that the object goes from right to left.
Usually the local time is modeled as: tau(t) =a* t + b, where the scalars a and b are called its skew (speed of clock) and offset (local time reading when t=0), respectively. A node can determine the global time t from its local clock reading by using the relationship t =(tau -b)/a as long as the node knows its skew and offset. Hence the problem of clock synchronization in a network can be posed as the problem of nodes estimating their skews and offsets with respect to a global reference time t. We address the problem of estimating clock skews and offsets in a distributed fashion in mobile sensor networks.
2. Occupancy estimation in commercial buildings
Occupancy estimation in commercial buildings are beneficial in multiple purpose such as energy saving, fire rescue and even business analysis (as shown in the picture above).
According to the 2008 building energy databook, in the United States, buildings are responsible for 38% of CO2 emissions, 71% of electricity consumption. Therefore, we are more interested in occupancy estimation as an input to the control of HVAC system in commercial buildings. As a result, we should look at the daily behavior of occupants, which is challenging.
Agent-based Model of Building Occupancy :
A model of occupancy dynamics can provide more realistic sample time-traces of occupancy over time from which, peaks, means and variances of occupancy driven heat gains can be computed. It is useful 1) in commissioning and re-commissioning of buildings and 2) as input to building energy simulation programs, e.g. EnergyPlus and ESP-r to compute loads.
We constructed and validated the agent-based model (MuMo model) in real building. The figure above is the floor plan of the 3rd floor of MAE-B building in the University of Florida campus. The whole floor is used for verification of the model in multiple-zone scenario and zone 15 is the room that is used for verification of single-zone scenario. The stars and triangle indicates the cameras using in the two scenarios respectively.
Single zone:
Multiple zones:
Occupancy estimation and prediction :
Due to the highly uncertain nature of occupancy and wide fluctuations seen over multiple time-scales, occupancy cannot be computed based on expected building use, and has to be estimated or predicted in real-time.
3. Energy Efficiency Building
I am currently involved in this project, please refer to this link.
Chenda Liao
About me :
- Ph.D. Candidate
- Distributed Control System Laboratory
- Department of Mechanical and Aerospace Engineering
- University of Florida
