Study On Remote Estimation Of One Kind Of Wireless Sensor Networks With Limited Energy

Wireless Sensor Networks(WSNs)have extensive applications in industrial and environmental monitoring.However,using WSNs to collect and transmit information of the surroundings needs to consume the energy of the sensor node,and once the en-ergy is depleted,the lifespan of the sensor node ends.This requires us to design energy scheduling strategies for the WSNs with limited energy.Scheduling strategies are often used to design routing paths and expand communication coverage,and in this disser-tation,the state estimation accuracy is used as the indicator of scheduling strategies.Therefore,although the low power signal transmission will increase the probability of the packet drop,we still need to make a trade-off between the transmission signal's qual-ity and power consumption.That means while under the energy constraint,we need to guarantee a certain life span of the system while trying to improve the accuracy of the state estimation,i.e.,the accuracy of the signal transmission.To sum up,in this dissertation,we discuss the trade-off between the estimation precision and the energy consumption for the transmission when the sensor is far away from the estimator.We use two models to bring down the packet drop rate and then improve the remote state estimation.First,we make state estimation for the WSN with relay nodes and search for the optimal sending strategy for the sensor and the relay nodes within limited time steps.Previous work adjusted the trade-off between the energy consumption and the estima-tion precision through a predefined threshold,while we use an algorithm to find the optimal sending strategy to obtain the minimum expected estimation error.The algo-rithm converts the sensor-to-relay-node way of estimation error covariance update into the relay-node-to-relay-node way,which enables us to compare the average error covari-ances of different sending strategies by majorization,so as to find the optimal strategy.The difficulty is to unify the different methods of error covariance updating so that the results of the updates can be compared with the majorization theory.Second,we use the WSN with a feedback loop and set up an online algorithm for the WSNs.Within limited time steps and limited energy budgets,our algorithm de-creases the estimation error of the estimator by deciding whether to send a high power signal or a low power signal.When there is no feedback loop,previous research has shown that the optimal transmission strategy should send the high power signals among the low power signals as uniformly as possible.While by receiving the feedback in-formation,the sensor knows whether a low-power signal has successfully reached the remote estimator or not.Those low power signals that have reached the estimator suc-cessfully can be regarded as high power signals,and these extra high power signals can be used to improve the state estimation.Therefore,we use the idea of predictive control,rolling optimization,and feedback correction in control theory to design algorithms that make our algorithm reschedule the sending strategy at every time step and obtains the minimum expected error covariance for the rest of the time steps.The difficulty lies in the feedback correction part.We will extend the method for optimal state estimation scheduling strategy from previous studies to find the strategies that optimize the feed-back correction and their sufficient conditions.We simulate our scheduling strategies under the two models and compare our strategies with other strategies.Finally,we summarize our scheduling strategies un-der these two models and discuss other models and strategies that we might apply in the future under the remote estimation scenario.