State Estimation and Task Distribution Optimization
in Sensor Networks Based on Cyclic Approach[1]
V. A. Erofeeva
St. Petersburg State University victoria@grenka.net
Key words: linear matrix inequalities,
stochastic optimization, parameter tracking, nonstationary functional,
multi-agent technologies, task distribution.
Due to significant advancements in embedded
systems, sensor devices and wireless communication technology, sensor networks
have attracted widespread attention in various areas of practical application.
Such networks are difficult to control because of uncertainties among which are
limited network throughput, measurement noise, asynchronous messaging, and
energy efficiency. Under uncertain conditions, the stochastic approximation
algorithms and linear matrix inequalities have proved to be very useful. In
this paper we propose an approach for solving the task distribution
optimization problem using linear matrix inequalities. Besides of that, we
generalize the cyclic stochastic approximation algorithm and obtain the
estimation properties. In addition, the cyclic stochastic approximation
algorithm is applied to the distributed state estimation problem.
Bibliogr.:
49 refs.
N. A. Kizhaeva
St. Petersburg State University natalia.kizhaeva@gmail.com
Key words: text mining, classification
algorithms, clustering algorithmns.
A method for dynamic modeling of texts is
introduced. Based on the proposed model two classification algorithms of text
documents are designed. First classification rule is based on clustering of
text documents using periodograms of times series constructed for each
document. Second classification rule applies clustering with kernel based
distances.
Bibliogr.:
32 refs.