Distributed Estimation Fusion and Control with Packet Losses for Industrial Wireless Sensor and Actuator Networks

Abstract

This paper is concerned with a reliable solution to the distributed estimation and control issues over hybrid two-tier industrial wireless sensor and actuator networks (IWSANs). It aims at applying wireless technology to industrial automation control domains in hash industrial environment. A main challenge in these application domains is that wireless communication channels must satisfy strict requirements of real time and reliability. Therefore, first, we propose a new interacting dual model (IDM) adaptive estimation algorithm to identify the unreliable wireless communication channels, and, second, to further enhance the accuracy and fault tolerance of state estimation for controlled plant, a federated multisensor estimation fusion approach is presented. Furthermore, considering that the IWSANs need commonly to be deployed over a vast geographical area, a distributed collaborative control strategy is adopted. Finally, an example of temperature control illustrates the effectiveness of the proposed method.

1. Introduction

Wireless communication represents a major industrial stake in the coming years following the Fieldbus and Industrial Ethernet [1]. With the increasing demand in spatially distributed and flexible industrial applications, automation enterprises have drawn great attention to the communication and control architectures based on industrial wireless sensor and actuator networks (IWSANs). The essential attribute of IWSANs brings several significant advantages compared with traditional wired counterparts from the perspective of flexibility and convenience of the deployment for sensors and actuators (in remote, dangerous, and hard-to-reach areas), low cost (due to the less demands for cables, materials, preinstalled infrastructure, maintenance, and labor), compositionality (where scalability of network can be carried out more easily), and so forth [2, 3]. The benefits of integrating wireless communication technology with industrial applications are compelling. Some exploratory work has been done such as drip irrigation control for agriculture using wireless sensor and actuator network (WSAN) [4], environment control system in WSAN [5, 6], wireless ventilation systems in mining industrial [7], wireless fire detection and control [8], towards usage of wireless MEMS in industrial context [9], and monitoring system for robots and inventories in assembly line based on wireless sensor networks (WSNs) [10].

However, a common outstanding problem in these aforementioned applications is that the unreliable wireless communication channels suffering from delays and losses of data are difficult to ensure real-time and reliability requirements of control systems. The problem hinders that wireless technology is accepted generally and permeated in many industrial application domains. Therefore, designing the communication protocol standards, network topology architecture, and appropriate estimation and control strategies over IWSANs for networked control systems (NCSs) in industrial applications are still open issues.

In current literatures, most commonly adopted models of unreliable wireless channels are I.I.D (independent and identically distributed) model and Gilbert-Elliott model. The I.I.D model describes packet losses by a I.I.D Bernoulli random variable (taking value 1 or 0), such as in [5]. Gilbert-Elliott model considers packet losses as a two-state Markov chain to describe packet arrival and drop behaviors in [11]. Based on the above-mentioned two models, many researches have got some significant results for the problems of estimation and control [12–20].

The fault-tolerant mechanism is also an important issue which is used to improve the reliability of wireless data transmission in hash industrial context. As the current research hot spot issue of signal processing area, multisensor information fusion [21] can improve the fault tolerance and accuracy of real-time field data by utilizing multiple spatially distributed sensors to provide complementary and overlapping coverage on plants. It has received significant attention. Xia et al. [22] consider the networked data-fusion system with packet losses. In [23–26], fault-tolerant fusion, identification, and control for wireless sensor networks are presented. Hierarchical sensor data aggregation/fusion is studied by [27, 28].

Currently, high-reliable estimation and control strategies for industrial NCSs are still challenging issues. Since the quality of state estimation of plant will have a significant impact on control performance, communication and control became closely related and thus cannot be considered independently. In this paper, we are committed to developing a distributed collaborative estimation and control strategy over IWSANs. The main contributions are summarized as follows. (1)

We use interacting dual model (IDM) adaptive estimation algorithm to identify online the unreliable communication channels within IWSANs. To the best of our knowledge, the problem formulation is novel. Further, a federated estimation fusion algorithm is presented to improve the performance of accuracy and fault tolerance of state estimation.

(2)

A distributed collaborative scheme is presented which is similar to [20] but is a direct coordination method among actuators without going through the sensors.

(3)

A hybrid two-tier IWSAN based on OCARI technology [1] is developed. Based on the network topology architecture, the presented distributed estimation and control methods in (1) and (2) can be readily “piggy-backed” into industrial wireless networks such as WIA-PA [29] and OCARI [30].

The rest of this paper is organized as follows. The control system model is introduced in Section 2. A new IDM estimation and federated fusion algorithm are stated in Section 3. We design the distributed collaborative control strategy in Section 4. A simulation example of temperature control is provided in Section 5. Finally, Section 6 concludes this paper.

Notations. $ℝ^{n}$ is the n-dimensional Euclidean space. $I_{n}$ denotes the $n \times n$ identity matrix. $P r {\cdot}$ , $p (\cdot)$ , and $𝔼 {\cdot}$ denote the operators for probability, probability density, and estimation, respectively. $| \cdot |$ denotes cardinality of a set. $A \to B$ denotes the map of set A to B. $T r (\cdot)$ denotes the trace of a matrix. $d i a g (\cdot)$ denotes a diagonal matrix.

2. Problem Formulation

In many industry fields such as printing, textile, refrigeration, and semiconductor, industrial workplaces put forwards strict requirements for some distinct parameters such as temperature, humidity, air cleanliness, pressure balance, and ventilation [5, 7]. In this paper, we consider an example of temperature supervision and control system in an industrial workshop. Based on OCARI technology, we place hybrid two-tier IWSANs organized in a cluster of cells in a workshop as shown in Figure 1. Each cell has a star topology and consists of an actuator (cell coordinator) with its group of sensors (end device nodes). The actuator is in charge of coordinating its sensors within cell and routing data packets to gateway. Furthermore, every actuator can exchange information with its neighbor actuators via the upper layer actuator network with mesh structure, so actuators can take decisions and collaboratively achieve a global target upon the industrial environment. In addition, the cooperation mechanism among actuators can improve the reliability of system. For example, when an actuator malfunctions, its neighbor actuators can be complements or alternate devices. In this paper, the actuator, also regarded as a processing center, can perform a multisource estimate fusion with its “radiofixed” intracell sensors through only single-hop communication channels. The single-hop star topology can overcome the high packet loss and long-time delays induced by multihop transmissions. The design of “radiofixed,” where the sensors need not choose their transmission paths to actuator, decreases the communication complexity and improves the real-time and reliability of data transmission.

Figure 1

Topology of hybrid two-tier IWSANs.

Our object is to control the temperature within every cell domain in a workshop to meet set value, respectively, according to the actual production requirements.

2.1. Graph of Hybrid Two-Tier IWSANs

In order to facilitate the following work, the communication relationship between the actuators and sensors and among actuators themselves over the hybrid two-tier IWSANs can be described by a graph 𝒢. Assume that there are n cells in a workshop. An actuator (i.e., central air conditioning) and a group of m temperature sensors with wireless communication function are deployed within every cell. At first, we focus on the graph of actuator network $𝒢_{a} ≜ {𝒜, ℰ}$ (i.e., a subgraph of 𝒢), where $𝒜 = {a_{1}, \dots, a_{n}}$ is the set of n actuators and

\begin{matrix} ℰ = {(a_{i}, a_{l}) ∣ a_{i}, a_{l} \in 𝒜, a_{l} \in 𝒩_{a_{i}}, \forall i, l \in {1, \dots, n}} \end{matrix}

(1)

is the edge set.

𝒩_{a_{i}}

is a set of the neighbor actuators of

a_{i}

. Each actuator is only allowed to communicate with neighbor actuators over actuator network, that is,

\exists (a_{i}, a_{l}) \in ℰ

, if and only if

a_{l} \in 𝒩_{a_{i}}

Consider that $a_{i}$ can communicate with a group of m intracell sensors (i.e., $𝒮_{i} = {s_{i}^{1}, \dots, s_{i}^{m}}$ ) over sensor network. Thus, the overall graph 𝒢 can be obtained via taking the subgraph $𝒢_{a}$ and adding $m \times n$ new vertices 𝒮, where $𝒮 = S_{1} \cup S_{2}, \dots, \cup S_{n}$ corresponds to the set of all sensors. Define the edge set

\begin{array}{l} ℰ_{𝒪} = {(s_{i}^{j}, a_{i}) ∣ a_{i} \in 𝒜, s_{i}^{j} \in 𝒮_{i}, \forall i \in {1, \dots, n}, \\ \forall j \in {1, \dots, m}} . \end{array}

(2)

Then, we obtain

𝒢 = {𝒜 \cup 𝒮, ℰ \cup ℰ_{𝒪}}

Remark 1.

In this paper, $ℰ_{𝒪}$ represents the radioconnectivity between actuators and their intracell sensors. However, ℰ represents not only the radio connectivity but also interactive acting among neighboring actuators through some ventilation pipes.

2.2. Modeling

Consider the above-mentioned temperature control system, where the temperature is controlled in an industrial workshop using hybrid two-tier IWSANs as presented in Figure 1. We focus on the following discrete-time plant model:

\begin{matrix} x_{i, k + 1} = g_{i} x_{i, k} + w_{i i} u_{i, k}^{a} + \sum_{a_{l} \in 𝒩_{a_{i}}} ‍ w_{i l} u_{l, k}^{a}, \\ y_{i, k} = h_{i} x_{i, k}, \end{matrix}

(3)

where

x_{i, k}

i \in {1, \dots, n}

is the temperature state of ith cell domain of plant at time k and is measured by a group of m intracell sensors, that is,

𝒮_{i} = {s_{i}^{1}, \dots, s_{i}^{m}}

y_{i, k} = {[y_{i, k}^{1}, \dots, y_{i, k}^{m}]}^{T}

is the generalized measurement value of

x_{i, k}

collected by

𝒮_{i}

and can be sent to actuator

a_{i}

, and

u_{i, k}^{a}

and

u_{l, k}^{a}

l \in {1, \dots, n}

, are the input (output wind temperature) corresponding to the acting applied to the state

x_{i, k}

by local actuator

a_{i}

and neighbor actuator

a_{l}

, respectively.

Due to the unreliable wireless communication channels and the interference from the harsh industrial environment, it is necessary to take into account a more “realistic” plant model. First, consider the unreliable wireless communication channels within actuator network. Let I.I.D Bernoulli stochastic variable $γ_{i l, k}$ with $𝔼 {γ_{i l, k}} = γ_{i l}$ indicate whether data packet (including state information) from $a_{i}$ is received ( $γ_{i l, k} = 1$ ) or lost ( $γ_{i l, k} = 0$ ) by neighbor $a_{l} \in 𝒩_{a_{i}}$ . Apparently, $γ_{i i, k} \equiv 1$ , for all $i \in {1, \dots, n}$ . After any neighbor $a_{l}$ of $a_{i}$ obtains the state information of $a_{i}$ , its control unit can compute an appropriate input to $a_{i}$ . Therefore, when $a_{l} \in 𝒩_{a_{i}}$ and $γ_{i l, k} = 1$ , $x_{i, k}$ will be actuated cooperatively by local actuator $a_{i}$ and its neighbor actuators ${a_{l}}$ . For instance, in Figure 1, except local $a_{1}$ , cell 1 domain can also receive the other inputs sent from $a_{2}$ , $a_{3}$ , and $a_{4}$ through some ventilation pipes, respectively.

For a concise derivation, according to the zero-input strategy in the literature [31], we can write

\begin{matrix} u_{i, k}^{a} = u_{i, k}, \\ u_{l, k}^{a} = u_{l, k}, \end{matrix}

(4)

where

u_{i, k}

and

u_{l, k}

are the desired control laws computed by embedded control units of

a_{i}

and

a_{l}

, respectively.

Secondly, we use the Gilbert-Elliott model to simulate the unreliable wireless channels within sensor network. At time k, the stochastic variable $θ_{i, k}$ is used to describe the packet loss model $θ_{i, k}^{0}$ and the packet arrival model $θ_{i, k}^{1}$ of communication channel from $𝒮_{i}$ to $a_{i}$ , for all $i \in {1, \dots, n}$ , where $θ_{i, k}^{α} ≜ {θ_{i, k} = α}$ , $α = 0,1$ . To capture the temporal correlation of channel variation, $θ_{i, k}$ can be modeled as a two-state Markovian chain with the initial model probability vector $π_{i, 0}$ and the transition probability matrix (TPM) $Ξ_{i}$ as

\begin{matrix} π_{i, 0} = [π_{i, 0}^{0} π_{i, 0}^{1}], \end{matrix}

(5)

\begin{matrix} Ξ_{i} = [\begin{bmatrix} ξ_{i}^{00} & ξ_{i}^{01} \\ ξ_{i}^{10} & ξ_{i}^{11} \end{bmatrix}], \end{matrix}

(6)

where

π_{i, 0}^{α} ≜ P r {θ_{i, 0}^{α}} (α = 0,1)

is the initial model probability,

ξ_{i}^{α β} ≜ P r {θ_{i, k}^{β} ∣ θ_{i, k - 1}^{α}} (α, β = 0,1)

is the transition probability of switching from model

θ_{i, k - 1}^{α}

to model

θ_{i, k}^{β}

, and

ξ_{i}^{00}

ξ_{i}^{11}

ξ_{i}^{01} = 1 - ξ_{i}^{00}

, and

ξ_{i}^{10} = 1 - ξ_{i}^{11}

are called packet remaining loss rate, packet remaining arrival rate, packet recovery rate, and packet failure rate, respectively. Apparently,

Ξ_{i}

, the TPM, is characterized by the

ξ_{i}^{00}

and

ξ_{i}^{11}

When aggregating all states at time k into the $x_{k} = {[x_{1, k}^{T}, \dots, x_{n, k}^{T}]}^{T}$ and considering packet losses, noises, and (4), the above plant model (3) can be written in vector form as

\begin{matrix} x_{k + 1} = G x_{k} + W (γ_{k}) u_{k} + ω_{k}, \\ y_{k} = H (θ_{k}) x_{k} + ν_{k}, \end{matrix}

(7)

where

x_{k} \in ℝ^{n}

u_{k} \in ℝ^{n}

y_{k} \in ℝ^{n m}

ω_{k} \in ℝ^{n}

ν_{k} \in ℝ^{n m}

ω_{k}

and

ν_{k}

are mutually uncorrelated Gaussian, white, and zero-mean noises with covariance

Q > 0

and covariance

R > 0

, respectively,

G = d i a g (g_{1}, \dots, g_{n})

W (γ_{k}) \in ℝ^{n \times n}

w_{i l} (γ_{i l, k}) = 0

a_{l} \notin 𝒩_{a_{i}}

γ_{i l, k} = 0

, and

H (θ_{k}) = d i a g (h_{1} (θ_{k}), \dots, h_{n} (θ_{k}))

with

h_{i} (θ_{k}) = [h_{i}^{1} (θ_{i, k}), \dots, h_{i}^{m} (θ_{i, k})]^{T}

In this paper, the main objective is to control temperature within every cell domain to a required set value. For the aforementioned stochastic system (7), firstly, let us give the following quadratic cost function:

\begin{matrix} J = \frac{1}{n} 𝔼 {{[ψ_{k} - x_{k})}^{T} (ψ_{k} - x_{k}]}, \end{matrix}

(8)

where

ψ_{k} = {[ψ_{1, k}, \dots, ψ_{n, k}]}^{T}

is the temperature set value vector. Then, by estimating the temperature state, users wish to find an admissible optimal control strategy

u_{k}^{*}

to minimize the cost function J as

\begin{matrix} u_{k}^{*} ≜ 𝔼_{y^{k}} {\underset{u_{k}}{argmin} J}, \end{matrix}

(9)

where

u_{k} \in u^{k}

u^{k} = {u_{τ} ∣ τ = 1, \dots, k}

, and

y^{k} = {y_{τ} ∣ τ = 1, \dots, k}

3. Estimation and Fusion

In this section, we present a new adaptive estimation and fusion algorithm based on IDM and federated filter (IDM-FF). The architecture of the IDM-FF is shown in Figure 2. The IDM-FF method can be divided into two phases: first, the single-sensor processing phase, where temperature of every cell domain is measured by m sensors, so the actuator provides m LFs (local filters) to preprocess the data received from the m sensors, respectively, that is, every LF runs an adaptive IDM filtering algorithm to identify online the unreliable communication channel in order to estimate the measurement data of sensors, and, second, the multisensor data fusion phase, where the MF (main filter) is only to fuse the data obtained from its m preprocessing LFs so as to improve the accuracy and fault tolerance against the harsh environments and electromagnetic interference issues in industrial workshop.

Figure 2

Architecture of the IDM-FF.

3.1. IDM Adaptive Filtering

As presented in Section 2.2, the unreliable wireless channels within sensor network can be modeled as a two-state Markovian chain. In many practical situations, however, the prior knowledge of TPM (6) about the channel may be inadequate and lacking especially for large-scale wireless networks. Therefore, we propose a new adaptive channel-aware method based on IDM filter to not only identify the channel status but also learn online a posterior estimation of TPM from collected data.

First, we give the adaptive learning process online to obtain the TPM without the prior knowledge; then the updated TPM can be stored and utilized in IDM filtering algorithm by LF for channel decoding. According to the Bayesian framework, the estimation ${\bar{Ξ}}_{i}$ of $Ξ_{i}$ , model probability $π_{i, k}$ , and model likelihood $Λ_{i, k}$ are defined, respectively, as follows:

\begin{array}{l} {\bar{Ξ}}_{i, k} ≜ 𝔼 [Ξ_{i} ∣ {(y_{i}^{j})}^{k}], \\ π_{i, k} = {[π_{i, k}^{0}, π_{i, k}^{1}]}^{T} with \\ π_{i, k}^{β} ≜ P r {θ_{i, k}^{β} ∣ {\bar{Ξ}}_{i, k - 1}, {(y_{i}^{j})}^{k - 1}}, β = 0,1, \\ \begin{array}{l} Λ_{i, k} = {[Λ_{i, k}^{0}, Λ_{i, k}^{1}]}^{T} with \end{array} \\ Λ_{i, k}^{β} ≜ p [y_{i, k}^{j} ∣ θ_{i, k}^{β}, {\bar{Ξ}}_{i, k - 1}, {(y_{i}^{j})}^{k - 1}], β = 0,1, \end{array}

(10)

where

i \in {1, \dots, n}

{(y_{i}^{j})}^{k} = {y_{i, τ}^{j} ∣ τ = 1, \dots, k}

At time k, we consider an adaptive recursive process to obtain the TPM according to posterior model probability $π_{i, k - 1}$ and likelihood $Λ_{i, k}$ .

By the whole probability formula, one can obtain

\begin{array}{l} p [y_{i, k}^{j} ∣ Ξ_{i}, {(y_{i}^{j})}^{k - 1}] \\ = \sum_{β = 0}^{1} ‍ p [y_{i, k}^{j} ∣ θ_{i, k}^{β}, Ξ_{i}, {(y_{i}^{j})}^{k - 1}] P r {θ_{i, k}^{β} ∣ Ξ_{i}, {(y_{i}^{j})}^{k - 1}} \\ = \sum_{β = 0}^{1} ‍ p [y_{i, k}^{j} ∣ θ_{i, k}^{β}, Ξ_{i}, {(y_{i}^{j})}^{k - 1}] \\ \times \sum_{α = 0}^{1} ‍ P r {θ_{i, k}^{β} ∣ θ_{i, k - 1}^{α}, Ξ_{i}, {(y_{i}^{j})}^{k - 1}} \\ \times P r {θ_{i, k - 1}^{α} ∣ Ξ_{i}, {(y_{i}^{j})}^{k - 1}} \\ \approx \sum_{β = 0}^{1} ‍ Λ_{i, k}^{β} \sum_{α = 0}^{1} ‍ ξ_{i}^{α β} π_{i, k - 1}^{α} \\ = π_{i, k - 1}^{T} Ξ_{i} Λ_{i, k}, \end{array}

(11)

where the following approximations

\begin{matrix} p [y_{i, k}^{j} ∣ θ_{i, k}^{β}, Ξ_{i}, {(y_{i}^{j})}^{k - 1}] \approx Λ_{i, k}^{β}, \\ P r {θ_{i, k - 1}^{α} ∣ Ξ_{i}, {(y_{i}^{j})}^{k - 1}} \approx π_{i, k - 1}^{α} \end{matrix}

(12)

are used in (11). These local linearity approximations are equivalent to replacing the unknown

Ξ_{i}

in (11) with its best estimate

{\bar{Ξ}}_{i, k - 1}

for the current recursive cycle

(k - 1) \to k

. This approximation approach can provide high accuracy as stated in [32].

According to (11), $p [y_{i, k}^{j} ∣ {(y_{i}^{j})}^{k - 1}]$ becomes

\begin{array}{l} p [y_{i, k}^{j} ∣ {(y_{i}^{j})}^{k - 1}] \\ = \int p [y_{i, k}^{j} ∣ Ξ_{i}, {(y_{i}^{j})}^{k - 1}] p [Ξ_{i} ∣ {(y_{i}^{j})}^{k - 1}] d Ξ_{i} \\ = \int π_{i, k - 1}^{T} Ξ_{i} Λ_{i, k} p [Ξ_{i} ∣ {(y_{i}^{j})}^{k - 1}] d Ξ_{i} \\ = π_{i, k - 1}^{T} {\bar{Ξ}}_{i, k - 1} Λ_{i, k} . \end{array}

(13)

Further, the approximate posterior probability density function (PDF) of TPM will be updated as follows:

\begin{matrix} p [Ξ_{i} ∣ {(y_{i}^{j})}^{k}] = \frac{p [y_{i, k}^{j} ∣ Ξ_{i}, {(y_{i}^{j})}^{k - 1}]}{p [y_{i, k}^{j} ∣ {(y_{i}^{j})}^{k - 1}]} p [Ξ_{i} ∣ {(y_{i}^{j})}^{k - 1}] . \end{matrix}

(14)

Substituting (11) and (13) into (14), we can obtain

\begin{matrix} p [Ξ_{i} ∣ {(y_{i}^{j})}^{k}] = \frac{π_{i, k - 1}^{T} Ξ_{i} Λ_{i, k}}{π_{i, k - 1}^{T} {\bar{Ξ}}_{i, k - 1} Λ_{i, k}} p [Ξ_{i} ∣ {(y_{i}^{j})}^{k - 1}] . \end{matrix}

(15)

Finally, we compute the estimation of TPM by a numerical integration as

\begin{array}{l} {\bar{Ξ}}_{i, k} = \int Ξ_{i} p [Ξ_{i} ∣ {(y_{i}^{j})}^{k}] d Ξ_{i} \\ = \frac{1}{N} \sum_{s = 1}^{N} ‍ Ξ_{i}^{s} p [Ξ_{i}^{s} ∣ {(y_{i}^{j})}^{k}], \end{array}

(16)

which means the posterior PDF distribution in N grid TPMs

Ξ_{i}^{s}

points over

[0,1]

Next, we use the IDM filter algorithm with ${\bar{Ξ}}_{i, k - 1}$ and $π_{i, k - 1}$ to estimate the measurement data. The basic idea of the presented IDM filtering approach requires running two same parallel optimal filters for the two possible channel model sequences (i.e., $θ_{i, k}^{0}$ and $θ_{i, k}^{1}$ ), and then the measurement data can be calculated using a combined estimate. We define the local mean and variance of $x_{i, k}^{j}$ measured and sent by sensor $s_{i}^{j}$ with the model $θ_{i, k}^{α}$ , $α = 0,1$ , at time k as follows:

\begin{matrix} {\hat{x}}_{i, k ∣ k}^{j, α} ≜ 𝔼 [x_{i, k}^{j, α} ∣ {(y_{i}^{j})}^{k}], \\ P_{i, k ∣ k}^{j, α} ≜ 𝔼 [(x_{i, k}^{j, α} - {\hat{x}}_{i, k ∣ k}^{j, α}) {(x_{i, k}^{j, α} - {\hat{x}}_{i, k ∣ k}^{j, α})}^{T} ∣ {(y_{i}^{j})}^{k}] . \end{matrix}

(17)

For brevity, we provide a summary of IDM filter algorithm as follows.

Step 1 (interacting for two models of varying channel).

The mixed probabilities $ξ_{i, k}^{β ∣ α}$ for channel models $θ_{i, k}^{α}$ and $θ_{i, k}^{β}$ at time k are calculated as

\begin{matrix} ξ_{i, k}^{β ∣ α} = \frac{{\bar{ξ}}_{i, k - 1}^{β α} π_{i, k - 1}^{β}}{\sum_{β = 0}^{1} ‍ {\bar{ξ}}_{i, k - 1}^{β α} π_{i, k - 1}^{β}}, \end{matrix}

(18)

where

π_{i, k - 1}^{β}

is probabilities for model

θ_{i, k}^{β}

at time

k - 1

{\bar{ξ}}_{k - 1}^{β α}

can be calculated and updated using (16) and is the recursively Bayesian posterior estimation of transition probabilities. Then, the mixed inputs (mean and covariance) for every filter can be computed as

\begin{array}{l} {\tilde{x}}_{i, k - 1}^{j, α} = \sum_{β = 0}^{1} ‍ ξ_{i, k}^{β ∣ α} {\hat{x}}_{i, k - 1 ∣ k - 1}^{j, β}, \\ {\tilde{P}}_{i, k - 1}^{j, α} = \sum_{β = 0}^{1} ‍ ξ_{i, k}^{β ∣ α} \\ \times {P_{i, k - 1 ∣ k - 1}^{j, β} + [{\hat{x}}_{i, k - 1 ∣ k - 1}^{j, β} - {\tilde{x}}_{i, k - 1}^{j, α}] \\ \times {[{\hat{x}}_{i, k - 1 ∣ k - 1}^{j, β} - {\tilde{x}}_{i, k - 1}^{j, α}]}^{T}}, \end{array}

(19)

where

{\hat{x}}_{i, k - 1 ∣ k - 1}^{j, β}

and

{\hat{P}}_{i, k - 1 ∣ k - 1}^{j, β}

are the updated state mean and covariance for model

θ_{i, k}^{β}

, respectively, at time

k - 1

Step 2 (information Kalman filler (IKF)).

For each model $θ_{i, k}^{α}$ , the IKF is done as $IK F_{u}$

\begin{array}{l} {\hat{x}}_{i, k ∣ k}^{j, α} = P_{i, k ∣ k}^{j, α} {(P_{i, k ∣ k - 1}^{j, α})}^{- 1} {\hat{x}}_{i, k ∣ k - 1}^{j, α} \\ + P_{i, k ∣ k}^{j, α} {(h_{i}^{j} (θ_{i, k}^{α}))}^{T} {(R_{i}^{j})}^{- 1} y_{i, k}^{j, α}, \\ {(P_{i, k ∣ k}^{j, α})}^{- 1} = {(P_{i, k ∣ k - 1}^{j, α})}^{- 1} \\ + {(h_{i}^{j} (θ_{i, k}^{α}))}^{T} {(R_{i}^{j})}^{- 1} h_{i}^{j} (θ_{i, k}^{α}), \end{array}

(20)

$IK F_{p}$ :

\begin{array}{l} {\hat{x}}_{i, k ∣ k - 1}^{j, α} = g_{i i} {\tilde{x}}_{i, k - 1}^{j, α} + w_{i i} u_{i, k - 1} \\ + \sum_{a_{l} \in 𝒩_{a_{i}}} ‍ w_{i l} (γ_{i l, k - 1}) u_{l, k - 1}, \\ {(P_{i, k ∣ k - 1}^{j, α})}^{- 1} = {(g_{i i} {\tilde{P}}_{i, k - 1}^{j, α} g_{i i}^{T} + Q_{i}^{j})}^{- 1}, \end{array}

(21)

where the prediction and update steps of IKF are denoted by $IK F_{p}$ and $IK F_{u}$ , respectively, $Q_{i}^{j} = β_{i}^{j} Q_{i}$ , according to the rule of information-distribution technology of federated filter stated in Section 3.2.

Step 3 (model probabilities update).

By Bayes’ rule, the probabilities of each model $θ_{i, k}^{α}$ at time k are updated as

\begin{matrix} π_{i, k}^{α} = \frac{(\sum_{β = 0}^{1} ‍ {\bar{ξ}}_{i, k - 1}^{β α} ξ_{i, k - 1}^{β}) Λ_{i, k}^{α}}{\sum_{α = 0}^{1} ‍ (\sum_{β = 0}^{1} ‍ ({\bar{ξ}}_{i, k - 1}^{β α} ξ_{i, k - 1}^{β}) Λ_{i, k}^{α})} . \end{matrix}

(22)

Step 4 (combined estimation).

The combined estimate for the state mean and covariance at time k are computed as

\begin{array}{l} {\hat{x}}_{i, k ∣ k}^{j} = \sum_{α = 0}^{1} ‍ π_{i, k}^{α} {\hat{x}}_{i, k ∣ k}^{j, α}, \\ P_{i, k ∣ k}^{j} = \sum_{α = 0}^{1} ‍ π_{i, k}^{α} \times {P_{i, k ∣ k}^{i, α} + [{\hat{x}}_{i, k ∣ k}^{j, α} - {\hat{x}}_{i, k ∣ k}^{j}] \\ \times {[{\hat{x}}_{i, k ∣ k}^{j, α} - {\hat{x}}_{i, k ∣ k}^{j}]}^{T}} . \end{array}

(23)

From the above results, we can know that the purpose of using IDM filter in every LF is to estimate the probability of channel model switching between packet loss model $θ_{i, k}^{0}$ and packet arrival model $θ_{i, k}^{1}$ to reduce the channel uncertainty and then obtain the combined estimation of state mean and covariance.

3.2. Federated Multisensor Estimation Fusion

In order to enhance the accuracy and fault tolerance of the whole measurement system in harsh industrial environment, the federated filter (FF) [33, 34], a decentralized filtering technology, is adopted to fuse the data of LFs (i.e., ${\hat{x}}_{i, k ∣ k}^{j}$ , $P_{i, k ∣ k}^{j}$ , $j = 1, \dots, m$ ) for every $x_{i, k}$ in MF.

To remove the correlation among the states in different LFs, according to the rule of information distribution, we have

\begin{matrix} Q_{i}^{j} = β_{i}^{j} Q_{i}, \end{matrix}

(24)

where

β_{i}^{j} > 0

is the information-distribution coefficient of the jth LF and is subjected to

\begin{matrix} \sum_{j = 1}^{m} ‍ β_{i}^{j} = 1 . \end{matrix}

(25)

Because

(h_{i}^{j})^{T} (R_{i}^{j})^{- 1} h_{i}^{j}

can reflect the measurement accuracy of local measurement,

σ_{i}^{j} = (h_{i}^{j})^{T} (R_{i}^{j})^{- 1} h_{i}^{j}

is defined as an information variable. Thus, the optimal information-distribution method [35] is given as

\begin{matrix} β_{i}^{j} = \frac{T r (σ_{i}^{j})}{\sum_{j = 1}^{m} ‍ T r (σ_{i}^{j})} . \end{matrix}

(26)

Then, the global state estimation ${\hat{x}}_{i, k ∣ k}^{g}$ and its covariance $P_{i, k ∣ k}^{g}$ obtained from the MF can be given as follows:

\begin{array}{l} P_{i, k ∣ k}^{g} = {[{(P_{i, k ∣ k}^{1})}^{- 1} + {(P_{i, k ∣ k}^{2})}^{- 1} + \dots + {(P_{i, k ∣ k}^{m})}^{- 1}]}^{- 1}, \\ {\hat{x}}_{i, k ∣ k}^{g} = P_{i, k ∣ k}^{g} [{(P_{i, k ∣ k}^{1})}^{- 1} {\hat{x}}_{i, k ∣ k}^{1} + {(P_{i, k ∣ k}^{2})}^{- 1} {\hat{x}}_{i, k ∣ k}^{2} \\ + \dots + {(P_{i, k ∣ k}^{m})}^{- 1} {\hat{x}}_{i, k ∣ k}^{m}], \end{array}

(27)

which indicates that the federated global estimation fusion

{\hat{x}}_{i, k ∣ k}^{g}

for

x_{i, k}

is a linear weighted combination of the local estimates

{\hat{x}}_{i, k ∣ k}^{j}

with weighting matrices

P_{i, k ∣ k}^{g} {(P_{i, k ∣ k}^{j})}^{- 1}

j = 1, \dots, m

. To a better fault tolerance of sensor measurements, we select a no-reset mode of federated filter which will not cross infect between LFs, as shown in Figure 2.

In summary, the main motivation for the above-mentioned results stems from the need to obtain reliable measurement data from sensors over wireless sensor network in the harsh industrial environment. In the following, we consider the cooperative control problem among actuators.

4. Distributed Control

In every cell domain within workshop, actuator can decide its control input according to the estimated local temperature state provided by its IDM-FF presented in Section 3. However, the stability of the whole system cannot be guaranteed due to overlapping actuation between actuators. Fortunately, the upper layer actuator network of two-tier IWSANs provides a foundation for coordinating the control inputs of actuators to improve the performance of control system. This implies that every actuator can obtain more complete state information of plant by exchanging information with neighbor actuators over the actuator network. Therefore, in this paper, we consider that every actuator can firstly perform a locally optimized control law without accounting for the effect of neighbor actuators, and then a cooperation mechanism is adopted to eliminate the overshoot caused by the interaction with neighbor actuators. Comparing the cooperation mechanism with the one presented in [20], the main difference between them is that any actuator can directly communicate with its neighbor actuators in our method, unlike the indirect communication between actuators going through the sensors in [20].

In this section, a locally optimal control method is considered firstly for actuator $a_{i}$ without collaborations with neighbor actuators: first, every $a_{i}$ can exchange information with neighbor $a_{l}$ if and only if $γ_{k} = 1$ via the upper actuator network. Second, every $a_{i}$ can decide its control input to its own ith cell domain and all neighbor l cell domains, respectively, based on the received information packets at the first step. Next, we describe the information flow among actuators by $𝒢_{a}$ . The input matrix $W (γ_{k})$ in (7) can be degenerated as a binary matrix $W^{'} (γ_{k})$ to indicate the connectivity of $𝒢_{a}$ by replacing any nonzero elements with one. Thus, we can use $W^{'} (γ_{k}) = [w_{i l}^{'} (γ_{i l, k})] \in ℝ^{n \times n}$ as the adjacency matrix of $𝒢_{a}$ :

\begin{matrix} w_{i l}^{'} (γ_{i l, k}) = {\begin{cases} 1 & if w_{i l} (γ_{i l, k}) \neq 0 \\ 0 & otherwrise . \end{cases} \end{matrix}

(28)

Remark 2.

Consider that $a_{i}$ is a neighborhood itself; that is, $w_{i i}^{'} (γ_{k}) \equiv 1$ in (28).

For $a_{i}$ , we define the number of its neighbors $𝒩_{a_{i}}$ as

\begin{matrix} ρ_{i, k} ≜ {| 𝒩_{a_{i}} |}_{γ_{k} = 1} = \sum_{a_{l} \in 𝒩_{a_{i}}} ‍ w_{i l}^{'} (γ_{k} = 1), \end{matrix}

(29)

where

1 ⩽ ρ_{i, k} ⩽ | 𝒩_{a_{i}} | ⩽ n

According to the aforementioned distributed collaborative control mechanism, every actuator needs to obtain the information of neighbor actuators before deciding a control law. So, $a_{i}$ can use a row selection matrix $Γ_{i, k} \in ℝ^{ρ_{i, k} \times n}$ to collect information from $ρ_{i, k}$ neighbor actuators. $Γ_{i, k}$ is defined as

\begin{matrix} Γ_{i, k} = {[e_{l}^{T}]}_{l ∣ w_{i l}^{'} (γ_{k}) = 1}, l = 1, \dots, n, \end{matrix}

(30)

where

e_{l}

denotes the lth vector of the standard basis of

ℝ^{n}

. Apparently,

Γ_{i, k}^{T}

is a column selection matrix. Thus, actuator

a_{i}

can receive available information from neighbor actuators (including itself) as

\begin{matrix} Ψ_{i, k} = Γ_{i, k} ψ_{k}, \end{matrix}

(31)

\begin{matrix} X_{i, k + 1} = G_{i, k} X_{i, k} + {\bar{W}}_{k} u_{k} + Ω_{i, k}, \end{matrix}

(32)

where

X_{i, k} = Γ_{i, k} x_{k}

Ω_{i, k} = Γ_{i, k} ω_{k}

with covariance

Q_{i}^{'} > 0

G_{i, k} = Γ_{i, k} G Γ_{i, k}^{T}

, and

{\bar{W}}_{k} = Γ_{i, k} W (γ_{k})

Remark 3.

Note that the process of selecting neighbor actuators of $a_{i}$ via $Γ_{i, k}$ is a dynamic process based on $γ_{k}$ due to the uncertainty of wireless actuator network. For example, $a_{l} \in 𝒩_{a_{i}}$ is a neighbor actuator of $a_{i}$ if $γ_{k - 1} = 1$ (packet reception) at time $k - 1$ but is not a neighbor actuator of $a_{i}$ if $γ_{k} = 0$ (packet loss) at time k.

From the aforementioned results, we can obtain the cost function only computed by $a_{i}$ as

\begin{matrix} J_{i, k} = \frac{1}{ρ_{i, k}} 𝔼 {{(Ψ_{i, k} - X_{i, k})}^{T} (Ψ_{i, k} - X_{i, k})} . \end{matrix}

(33)

Theorem 4.

At time $k - 1$ , one can find the locally optimal control law $u_{i, k}^{*}$ of actuator $a_{i}$ without accounting for the effects of the neighbor actuators $𝒩_{a_{i}}$ as follows:

\begin{array}{l} u_{i, k}^{*} = 𝔼 {\underset{u_{i, k}}{argmin} J_{i, k}} \\ = {({\bar{W}}_{i, k}^{T} {\bar{W}}_{i, k})}^{- 1} {\bar{W}}_{i, k}^{T} (Ψ_{i, k + 1} - G_{i, k} {\hat{X}}_{i, k ∣ k}), \end{array}

(34)

where

{\hat{X}}_{i, k ∣ k} = Γ_{i, k} [{\hat{x}}_{1, k ∣ k}^{g}, \dots, {\hat{x}}_{n, k ∣ k}^{g}]^{T}

, and

{\bar{W}}_{i, k}

is the ith column of matrix

{\bar{W}}_{k}

Proof.

According to (33), one can obtain

\begin{array}{l} J_{i, k + 1} = \frac{1}{ρ_{i, k + 1}} 𝔼 {{(Ψ_{i, k + 1} - X_{i, k + 1})}^{T} (Ψ_{i, k + 1} - X_{i, k + 1})} \\ = \frac{1}{ρ_{i, k + 1}} (𝔼 {- 2 Ψ_{i, k + 1}^{T} X_{i, k + 1} + X_{i, k + 1}^{T} X_{i, k + 1}} \\ + Ψ_{i, k + 1}^{T} Ψ_{i, k + 1}) . \end{array}

(35)

Substituting (32) into (35), we have

\begin{array}{l} J_{i, k + 1} \\ = \frac{1}{ρ_{i, k + 1}} (𝔼 {- 2 Ψ_{i, k + 1}^{T} (G_{i, k} X_{i, k} + {\bar{W}}_{k} u_{k} + Ω_{i, k}) \\ + {(G_{i, k} X_{i, k} + {\bar{W}}_{k} u_{k} + Ω_{i, k})}^{T} \\ \times (G_{i, k} X_{i, k} + {\bar{W}}_{k} u_{k} + Ω_{i, k})} + Ψ_{i, k + 1}^{T} Ψ_{i, k + 1}) \\ = \frac{1}{ρ_{i, k + 1}} (𝔼 {u_{k}^{T} {\bar{W}}_{k}^{T} {\bar{W}}_{k} u_{k} + 2 u_{k}^{T} {\bar{W}}_{k}^{T} \\ \times (G_{i, k} X_{i, k} - Ψ_{i, k + 1})}  +  Ψ_{i, k + 1}^{T} Ψ_{i, k + 1} + T r (Q_{i}^{'}) \\ + 𝔼 {X_{i, k}^{T} G_{i, k}^{T} G_{i, k} X_{i, k} - 2 X_{i, k}^{T} G_{i, k}^{T} Ψ_{i, k + 1}}) . \end{array}

(36)

The minimum of

J_{i, k + 1}

in (36) controlled by

u_{k}

\begin{array}{l} V_{i, k + 1} = \frac{1}{ρ_{i, k + 1}} (\min_{u_{k}} 𝔼 {u_{k}^{T} {\bar{W}}_{k}^{T} {\bar{W}}_{k} u_{k} \\ + 2 u_{k}^{T} {\bar{W}}_{k}^{T} (G_{i, k} X_{i, k} - Ψ_{i, k + 1})}) . \end{array}

(37)

By solving

\partial V_{i, k + 1} / \partial u_{k} = 0

, that is,

\begin{matrix} \frac{\partial V_{i, k + 1}}{\partial u_{k}} = 2 {\bar{W}}_{k}^{T} {\bar{W}}_{k} u_{k} + 2 {\bar{W}}_{k}^{T} (G_{i, k} {\hat{X}}_{i, k ∣ k} - Ψ_{i, k + 1}) = 0, \end{matrix}

(38)

where

{\hat{X}}_{i, k ∣ k} = Γ_{i, k} {[{\hat{x}}_{1, k ∣ k}^{g}, \dots, {\hat{x}}_{n, k ∣ k}^{g}]}^{T}

, we can obtain the optimal control law

u_{i, k}^{*}

\begin{matrix} u_{i, k}^{*} = {({\bar{W}}_{k}^{T} {\bar{W}}_{k})}^{- 1} {\bar{W}}_{k}^{T} (Ψ_{i, k + 1} - G_{i, k} {\hat{X}}_{i, k ∣ k}) . \end{matrix}

(39)

Since $u_{1, k}, u_{2, k}, \dots, u_{n, k}$ are independent, the locally optimal control law $u_{i, k}^{*}$ of $a_{i}$ without accounting for the effects of the neighbor actuators $𝒩_{a_{i}}$ can be expressed in (34). This completes the proof.

The locally optimal control law $u_{i, k}^{*}$ stated in Theorem 4 does not consider the issues of cooperation among neighbor actuators for the stability of the entire system. Therefore, in order to eliminate the overlapping action between neighbor actuators, we adopt a distributed cooperation mechanism for every actuator as follows:

\begin{matrix} u_{i, k}^{†} = μ_{i, k} u_{i, k}^{*}, i = 1, \dots, n, \end{matrix}

(40)

where

μ_{i, k}

is a tuning parameter. Next, we introduce how to compute

μ_{i, k}

for

a_{i}

. First, at time k, every actuator

a_{l}

a_{l} \in 𝒩_{a_{i}}

, can use an information set

ℐ_{l, k} = {w_{l j} (γ_{l j, k}) u_{j, k}^{*} ∣ a_{j} \in 𝒩_{a_{l}}, γ_{l j, k} = 1}

, where

u_{j, k}^{*}

is calculated by (34). Second, every

a_{l}

computes the number of elements in set

ℐ_{l, k}

that have the same sign as one another and then obtains a factor set

Λ = {λ_{j l, k}}

. For

a_{l}

λ_{j l, k}

means the number of its neighbor actuators that can produce overlapping actuation with

a_{l}

at time k. Then, after receiving

Λ

, we can obtain the tuning parameter for

a_{i}

\begin{matrix} μ_{i, k} = \frac{1}{ρ_{i, k}} \sum_{a_{l} \in N_{a_{i}}} ‍ \frac{1}{λ_{i l, k}} . \end{matrix}

(41)

Note that the aforementioned cooperation mechanism in (41) is similar to [20]; however, our method runs a direct cooperation among actuators without going through the sensors.

5. Numerical Example

In this section, we consider an example of temperature control in industrial environment over the hybrid two-tier IWSANs. The simulation experiment is built on MATLAB software platform. The monitored workshop field is 100 m × 100 m and deployed 7 cells. The actuating radius of actuators within each cell is $r_{a} = 17$ m, and the detection radius of each sensor is $r_{s} = 8$ m. Assume that the environment temperature of every cell domain can be controlled by a central air conditioning (i.e., an actuator with estimate/control unit) and collected by 5 intracell sensors. According to the actual production requirements, assume that our object is to control the temperature within every cell domain to meet set value $Ψ_{i}$ , $i \in {1, \dots, 7}$ , chosen from $[10^{°} C, 40^{°} C]$ . Let the plant model be specified as follows. Consider $g_{i i} = 0.9$ , for all i, and weight coefficient $w_{i l} (γ_{i l, k})$ is distributed randomly in $[0,1]$ . The average loss packet rate of channel is $𝔼 (γ_{i l, k}) = 75 %$ in the actuator network. For the sensor network, assume that $θ_{1, k} = θ_{2, k} = \dots = θ_{7, k} = θ_{k}$ , $h_{i}^{j} (θ_{k}^{α}) = α$ , $α = 0,1$ , the initial channel model probability is $π_{0} = [0.5 0.5]$ , the true TPM is chosen with $ξ^{00} = 0.3$ and $ξ^{11} = 0.9$ , and the IDM uses a numerical integration estimator with $50$ grid points over $[0,1]$ for posterior TPM.

In simulation running, let the number of the total simulation steps and the measurement packets from sensors successfully arriving at actuator be Ϝ and $Δ$ (i.e., the number of model $θ_{k}^{1}$ ), respectively. According to the TPM in (6), the times channel state $θ_{k}$ switching from $θ_{k}^{1}$ to $θ_{k}^{0}$ is about $Δ ξ^{10}$ . We introduce a variable $T_{loss} \in ℤ$ that denotes the times of $θ_{k}$ remaining at $θ_{k}^{0}$ . Therefore, the average times $θ_{k}$ remaining at $θ_{k}^{0}$ can be estimated as

\begin{matrix} 𝔼 {T_{loss}} = \sum_{t = 1}^{\infty} ‍ t {(ξ^{00})}^{t - 1} = \frac{1}{{(1 - ξ^{00})}^{2}} = \frac{1}{{(ξ^{01})}^{2}} . \end{matrix}

(42)

The total times

θ_{k}^{0}

in sequence of

θ_{k}

can be described approximately as

Δ ξ^{10} / {(ξ^{01})}^{2}

; thus

\begin{matrix} Δ + Δ \frac{ξ^{10}}{{(ξ^{01})}^{2}} \approx Ϝ . \end{matrix}

(43)

Finally, the above equation can be rewritten as

\begin{matrix} Δ \approx \frac{{(ξ^{01})}^{2}}{ξ^{10} + {(ξ^{01})}^{2}} Ϝ . \end{matrix}

(44)

If we take $Ϝ = 100$ , then $Δ \approx 83$ , $Δ ξ^{10} \approx 9$ , and the average arrival rates of channel state $θ_{k}$ are about $83 %$ . We report the approximate distribution of $T_{loss}$ in Table 1. A sample sequence of $θ_{k}$ is generated as shown in Figure 3.

Table 1

Analysis of packet losses.

Distribution of $T_{loss}$
$T_{loss}$	$P r {T_{loss} = t} = ξ^{01} (ξ^{00})^{t - 1}$	Times of $T_{loss} = t$
1	0.7000	5
2	0.2100	2
3	0.0630	1
4	0.0189	1

Figure 3

Example of sample sequence of $θ_{k}$ .

The results of simulation are shown in Figures 4–8. Figure 4 illustrates the control performance of the system. It is clearly seen that the system with cooperation is stable, but the cost function curve without cooperation has an about $20 %$ deviation due to overlapping effects among neighbor actuators. Figure 5 illustrates the convergence of the TPM estimators. Figure 6 shows that the adaptive IDM has a better accuracy of state estimate than the nonadaptive one with a mismatched TPM ( $ξ^{00} = ξ^{11} = 0.6$ ). Because the prior knowledge of true TPM for wireless communication channel with random packet losses is unknown in most real situations, it is preferable to use an adaptive algorithm to estimate the TPM. Figure 7 compares the accuracy of mean trace for estimation covariance for LF ( $Tr (p^{l})$ ), SF ( $Tr (p^{s})$ ), and MF ( $Tr (p^{g})$ ). Note that SF (single filter) indicates this case in which the temperature state $x_{i}$ is measured by single sensor rather than a multisensor measurement of federated estimation fusion. It shows that, first, the accuracy of MF is improved much compared with LF, and, second, federated estimators perform better than estimators without fusion. Figure 8 shows that the adaptive IDM filtering can provide a better true model probability. As one can expect, the distributed collaborative control approach can ameliorate effectively the control performance of the whole system, and the adaptive IDM-FF shows substantially better overall performance than the nonadaptive one.

Figure 4

Control performance.

Figure 5

Convergence of TPM estimators.

Figure 6

Mean absolute estimate error.

Figure 7

Mean trace for estimate covariance.

Figure 8

IDM probability of the true model.

6. Conclusions

In this paper, we consider the problem of distributed control and estimate over hybrid two-tier IWASNs for industrial automation control applications. In order to identify the unreliable communication channel, a novel IDM with the adaptive channel-aware algorithm is designed. The federated estimation fusion algorithm is presented to further improve the performance of accuracy and fault tolerance of state estimation. A distributed control scheme with coordination is proposed. The system stability and effectiveness of the presented methods are shown by simulation results.

Future work includes finding a multi-index (such as temperature, humidity, and air cleanliness) fusion and control strategy for production environments in a workshop and considering a coordination mechanism to minimize the average energy consumption of system.

Footnotes

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (61174070), the Specialized Research Fund for the Doctoral Program (20110172110033), and the Major Project for University-Industry Cooperation in Science and Technology of Fujian Province (2011H6023).

References

Al Agha

Bertin

M.-H.

Dang

Guitton

Minet

Val

Viollet

J.-B.

Which wireless technology for industrial wireless sensor networks? The development of OCARI technology

IEEE Transactions on Industrial Electronics 2009 56 10 4266 4278

2-s2.0-70349612965

10.1109/TIE.2009.2027253

Pajic

Sundaram

Pappas

G. J.

Mangharam

The wireless control network: a new approach for control over networks

IEEE Transactions on Automatic Control 2011 56 10 2305 2318

2-s2.0-80053628504

10.1109/TAC.2011.2163864

Akyildiz

I. F.

Sankarasubramaniam

Cayirci

Wireless sensor networks: a survey

Computer Networks 2002 38 4 393 422

2-s2.0-0037086890

10.1016/S1389-1286(01)00302-4

Wang

Hong

T. S.

Franzen

J. N.

Closed-loop drip irrigation control using a hybrid wireless sensor and actuator network

Science China Information Sciences 2011 54 3 577 588

2-s2.0-79952444230

10.1007/s11432-010-4086-6

Cao

Chen

Xiao

Sun

Building-environment control with wireless sensor and actuator networks: centralized versus distributed

IEEE Transactions on Industrial Electronics 2010 57 11 3596 3605

2-s2.0-77957994676

10.1109/TIE.2009.2029585

Nakamura

Sakurai

Furubo

Ban

Collaborative processing in Mote-based sensor/actuator networks for environment control application

Signal Processing 2008 88 7 1827 1838

2-s2.0-41249093661

10.1016/j.sigpro.2008.01.025

Witrant

D'Innocenzo

Sandou

Santucci

Di Benedetto

M. D.

Isaksson

A. J.

Johansson

K. H.

Niculescu

S.-I.

Olaru

Serra

Tennina

Tiberi

Wireless ventilation control for large-scale systems: the mining industrial case

International Journal of Robust and Nonlinear Control 2010 20 2 226 251

2-s2.0-75749101387

10.1002/rnc.1485

Akyildiz

I. F.

Kasimoglu

I. H.

Wireless sensor and actor networks: research challenges

Ad Hoc Networks 2004 2 4 351 367

2-s2.0-5444275208

10.1016/j.adhoc.2004.04.003

Minet

Bourgeois

Towards usage of wireless MEMS Networks in Industrial context

Proceedings of the 2nd Workshop on Design, Control and Software Implementation for Distributed MEMS (dMEMS '12)

2012

58 65

10.

Lee

Kwon

Song

Group connectivity model for industrial wireless sensor networks

IEEE Transactions on Industrial Electronics 2010 57 5 1835 1844

2-s2.0-77951177583

10.1109/TIE.2009.2033089

11.

Huang

Dey

Stability of Kalman filtering with Markovian packet losses

Automatica 2007 43 4 598 607

2-s2.0-33847417432

10.1016/j.automatica.2006.10.023

12.

Imer

O. C.

Yüksel

Başar

Optimal control of LTI systems over unreliable communication links

Automatica 2006 42 9 1429 1439

2-s2.0-33746037180

10.1016/j.automatica.2006.03.011

13.

Garone

Sinopoli

Goldsmith

Casavola

LQG control for MIMO systems over multiple erasure channels with perfect acknowledgment

IEEE Transactions on Automatic Control 2012 57 2 450 456

2-s2.0-84856477768

10.1109/TAC.2011.2167789

14.

Schenato

Sinopoli

Franceschetti

Poolla

Sastry

S. S.

Foundations of control and estimation over lossy networks

Proceedings of the IEEE 2007 95 1 163 187

2-s2.0-64149120662

10.1109/JPROC.2006.887306

15.

Shi

Epstein

Murray

R. M.

Kalman filtering over a packet-dropping network: a probabilistic perspective

IEEE Transactions on Automatic Control 2010 55 3 594 604

2-s2.0-77949422929

10.1109/TAC.2009.2039236

16.

Liu

Filter designing with finite packet losses and its application for stochastic systems

IET Control Theory & Applications 2011 5 6 775 784

10.1049/iet-cta.2010.0178

17.

Liu

Feng

Optimal filters with multiple packet losses and its application in wireless sensor networks

Sensors 2010 10 4 3330 3350

2-s2.0-77951743043

10.3390/s100403330

18.

Yan

L. P.

Zhou

D. H.

M. Y.

Xia

Y. Q.

State estimation for asynchronous multirate multisensor dynamic systems with missing measurements

IET Signal Processing 2010 4 6 728 739

2-s2.0-78649921474

10.1049/iet-spr.2009.0215

19.

Liu

B. S.

Yan

L. P.

Shi

State fusion estimation with missing measurements

Proceedings of the International Conference on Information Computing and Automation

2008

899 902

20.

Chen

Cao

Cheng

Xiao

Sun

Distributed collaborative control for industrial automation with wireless sensor and actuator networks

IEEE Transactions on Industrial Electronics 2010 57 12 4219 4230

2-s2.0-78349233534

10.1109/TIE.2010.2043038

21.

Hall

D. L.

Mathematical Techniques in Multisensory Data Fusion 1992

Boston, Mass, USA

Artech House

22.

Xia

Shang

Chen

Liu

G.-P.

Networked data fusion with packet losses and variable delays

IEEE Transactions on Systems, Man, and Cybernetics B 2009 39 5 1107 1120

2-s2.0-67449090817

10.1109/TSMCB.2009.2012437

23.

Hereford

J. M.

Fault-tolerant sensor systems using evolvable hardware

IEEE Transactions on Instrumentation and Measurement 2006 55 3 846 853

2-s2.0-33744519129

10.1109/TIM.2006.873791

24.

Krishnamachari

Iyengar

Distributed Bayesian algorithms for fault-tolerant event region detection in wireless sensor networks

IEEE Transactions on Computers 2004 53 3 241 250

2-s2.0-1842478856

10.1109/TC.2004.1261832

25.

Rongjun

Siyuan

Naigang

Practical multi-fading two-stage fault-tolerant federated filter for multi-sensor information fusion

Proceedings of the 6th World Congress on Intelligent Control and Automation (WCICA '06)

June 2006

5600 5604

2-s2.0-34047202244

10.1109/WCICA.2006.1714146

26.

Mao

Jiang

Fault identification and fault-tolerant control for a class of networked control systems

International Journal of Innovative Computing, Information and Control 2007 3 5 1121 1130

2-s2.0-41849122894

27.

Sridhar

Madni

A. M.

Jamshidi

Hierarchical aggregation and intelligent monitoring and control in fault-tolerant wireless sensor networks

IEEE Systems Journal 2007 1 1 38 54

28.

Kähler

Denzler

Triesch

Hierarchical sensor data fusion by probabilistic cue integration for robust 3-D object tracking

Proceedings of the IEEE Southwest Symposium on Image Analysis and Interpretation

March 2004

216 220

2-s2.0-2942553895

29.

Liang

Zhang

Xiao

Wang

Zeng

Survey and experiments of WIA-PA specification of industrial wireless network

Wireless Communications and Mobile Computing 2011 11 8 1197 1212

2-s2.0-80051770988

10.1002/wcm.976

30.

Dang

Devic

OCARI: optimization of communication for ad hoc reliable industrial networks

Proceedings of the 6th IEEE International Conference on Industrial Informatics

July 2008

688 693

2-s2.0-54849425153

10.1109/INDIN.2008.4618189

31.

Schenato

To zero or to hold control inputs with lossy links?

IEEE Transactions on Automatic Control 2009 54 5 1093 1099

2-s2.0-67349236204

10.1109/TAC.2008.2010999

32.

Jilkov

V. P.

X. R.

Online Bayesian estimation of transition probabilities for Markovian jump systems

IEEE Transactions on Signal Processing 2004 52 6 1620 1630

2-s2.0-2942619107

10.1109/TSP.2004.827145

33.

Calson

N. A.

Information-sharing approach to federated kalman filtering

Proceedings of the National Aerospace and Electronics Conference (NAECON '88)

1988

34.

Calson

N. A.

Federated aquare root filter for decentralized parallel processes

IEEE Transactions on Aerospace and Electronic Systems 1990 26 3 517 525

35.

Chen

T.-R.

Qiu

Pan

Optimal information sharing algorithm for the federated information filter without feedback

Chinese Journal of Sensors and Actuators 2007 20 5 1064 1067

2-s2.0-34547770714