Optimal linear quadratic control for wireless sensor ... - SAGE Journals

Research Article

Optimal linear quadratic control for wireless sensor and actuator networks with random delays and packet dropouts

International Journal of Distributed Sensor Networks 2018, Vol. 14(6) Ó The Author(s) 2018 DOI: 10.1177/1550147718779560 journals.sagepub.com/home/dsn

Zhuwei Wang1,2,3, Lihan Liu1,4, Chao Fang1,2,3, Xiaodong Wang5, Pengbo Si1,2,3 and Hong Wu4

Abstract In this article, the optimal linear quadratic control problem is considered for the wireless sensor and actuator network with stochastic network-induced delays and packet dropouts. Considering the event-driven relay nodes, the optimal solution is obtained, which is a function of the current plant state and all past control signals. It is shown that the optimal control law is the same for all locations of the controller placement. Since the perfect plant state information is available at the sensor, the optimal controller should be collocated with the sensor. In addition, some issues such as the plant state noise and suboptimal solution are also discussed. The performance of the proposed scheme is investigated by an application of the load frequency control system in power grid. Keywords Networked control systems, wireless sensor and actuator network, decentralized control, optimal control law

Date received: 27 November 2016; accepted: 23 April 2018 Handling Editor: Antonio Lazaro

Introduction Networked control systems (NCSs) in which the shared communication medium is used for the connection between the plant and the controller have recently attracted much attention due to their potential applications in various areas such as power grids, dc motors, and robotic networks.1,2 In such systems, the networkinduced delay and the packet dropout introduced by the communication network cause performance degradation and potential system instability. In the literature, network-induced delays have been modeled in various forms such as constant delays,3 short stochastic delays,4 and long stochastic delays.5 The typical approaches to modeling, analysis, and synthesis of NCSs with delays are summarized in Hespanha et al.6 and Zhang et al.7 On the other hand, the packet dropout can be modeled either as a

stochastic process8 or a deterministic one.3 If the packet containing the current state information is dropped, the typical solution is to either predict the lost state,9 or

1

Beijing Advanced Innovation Center for Future Internet Technology, Beijing University of Technology, Beijing, P.R. China 2 School of Information and Communications Engineering, Beijing University of Technology, Beijing, P.R. China 3 Beijing Laboratory of Advanced Information Networks, Beijing University of Technology, Beijing, P.R. China 4 Beijing University of Posts and Telecommunications, Beijing, P.R. China 5 Electrical Engineering Department, Columbia University, New York, USA Corresponding author: Chao Fang, School of Information and Communications Engineering, Beijing University of Technology, No.100 Ping Le Yuan, Chaoyang District, Beijing 100124, P.R. China. Email: [email protected]

Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/ open-access-at-sage).

2 use the previous state values,3 or simply set the lost state to zero.10 With full state information, the optimal control problem can be solved. In particular, the optimal linear quadratic (LQ) controller is developed for the NCS in Nilsson et al.4 and Hu and Zhu5 with stochastic network-induced delays. Compared with the traditional NCS, wireless NCSs, especially wireless sensor and actuator networks (WSANs), offer architectural flexibility and additional degrees of freedom.11 Several new standards have recently been introduced for multi-hop WSANs, for example, WirelessHART,12 ISA-100.13 Due to the flexibility of placing the controller in multi-hop WSANs, two significant challenges arise: the optimal placement of the controller node and the optimal control law design, which have been studied by a significant amount of works on WSANs. In particular, assuming the packets to be arbitrarily long, the problem of optimal controller placement in multi-hop WSANs is addressed in Robinson and Kumar,14 and it is shown that placing the controller at the actuator node achieves better performance than at the sensor node under certain conditions. Over the lossy networks with packet dropouts, the optimal control problem is decomposed into an LQ optimal regulator design and state estimation problem in Gupta et al.,15 and then the optimal control algorithm is obtained by a backward recursion for every time step. Assuming the sensor and the actuator can exchange data with the controller independently, the LQ Gaussian optimal control problem is addressed in Garone et al.16 for both finite and infinite horizon cases. The LQ Gaussian problem has also been studied in Lin et al.17 for systems with packet losses but without acknowledgment signals of status of control packet losses. In Xin et al.,18 the necessary and sufficient conditions under which the closed-loop WSAN is guaranteed to be stable for arbitrary controller location are studied in both one-hop and multi-hop network cases. Given the predesigned control strategy, the location distribution for the optimal controller placement is found in Quevedo et al.,19 where the actuator node is assumed to have the capability to broadcast the plant state information to the intermediate nodes over parallel unreliable links. However, only the effect of the packet dropout is considered in all the above works, and in this case the optimal control law is linear with the plant state (or estimated state). On the other hand, in practice, the network-induced delay and the packet dropout are the two most important characteristics that should be jointly studied in wireless NCSs.6,7 Taking the delay into consideration, not only the plant state but also the previous control signals should be used in the design of the optimal controller.4,5,20,21 In this article, we focus on the general case in which both the stochastic network-induced delays and the packet dropouts are considered, and the optimal

International Journal of Distributed Sensor Networks control problem is investigated for a linear WSAN with event-driven relay nodes. Using the quadratic cost function, the optimal solution is derived as a function that is linear with the current plant state and all past control signals, and the corresponding suboptimal solution is also proposed. Then, the optimal controller location is deserved to be at the sensor since the perfect plant state is available. The performance of the proposed algorithm is assessed using the application of load frequency control (LFC) in power grid systems. The remainder of this article is organized as follows. The system model and problem formulation are given in section ‘‘System model and problem formulation.’’ We then derive the optimal controller with stochastic delays and packet dropouts in section ‘‘Optimal control law design.’’ Section ‘‘Extension to the long delay case’’ extends the results to the case of long stochastic network-induced delays. Numerical results and conclusions are given in sections ‘‘Simulation results’’ and ‘‘Conclusion,’’ respectively.

System model and problem formulation In general, one sensor (or relay) node is selected as the controller to maintain the system stability in WSANs. As shown in Figure 1, node 5 is selected as the controller to coordinately generate control signals which are fed back to the plant. Actually, the WSAN considered in Figure 1 can be simplify converted to be the equivalent wireless multi-hop WSAN as shown in Figure 2,14,19 where the plant, sensor, actuator, and a number of relay nodes together form a closed-loop NCS. We assume that the plant is a continuous-time linear time-invariant system, the sensor is time-driven with a constant sampling period T , while the relay nodes and the actuator are event-driven. At each sampling instant, the measurement of the plant state is sent to the actuator through the multi-hop wireless network, and the controller can be located at either one of the sensor, relay nodes, or actuator to maintain the system stability.14,15,18,19 The transmission in the WSAN always

Figure 1. The structure of the WSAN.

Wang et al.

3

Figure 2. The wireless multi-hop WSAN.

suffers network-induced delays and packet dropouts. For a given transmission in the l-th hop, that is, from node l to node l + 1, we denote the network-induced delay tl and the packet dropout ul , which is a random switch such that ul = 1 indicates that the packet is successfully transmitted, whereas ul = 0 indicates that the packet is lost. Here, we consider a general case that both network-induced delays and packet dropouts are taken into consideration, and all of them are assumed to be independent and stochastic, such that only their distributions are known.4,5,18,19 For a given controller located at the lth node, the sensor-to-controller (S/C) delay tsc l , the controller-toactuator (C/A) delay tca , and the sensor-to-actuator (S/ l A) delay ttot are given respectively as tsc l = t

tot

l1 P i=1

ti , tca l =

ca = tsc l + tl

=

0 MP 1

ti ,

i=l 0 MP 1

ð1Þ

ti

i=1

where M 0 is the number of nodes through the sensorto-actuator transmission and the S/C, C/A, and S/A ca tot are given respectively packet dropouts usc l , ul , and u as usc l =

l1 Q i=1

ui , uca l =

0 MQ 1

ui ,

i=l

0

ca utot = usc l ul =

MQ 1

ð2Þ

ui

i=1

Note that, from equations (1) and (2), we can see that the S/A delay and packet dropout are independent of the controller node location, that is, unrelated to the parameter l. In this section, we first assume that the S/A delay is smaller than one sampling period, that is, ttot  T , and then the results are extended to the large delay case in section ‘‘Extension to the long delay case.’’ Considering the network-induced delay, the dynamic of the controlled plant is given by the following linear continuous-time model x_ (t) = Ax(t) + Bual ðt  ttot Þ

ð3Þ

where x(t) is an M-dimensional plant state vector, ual ðtÞ is the K-dimensional actuator input, and A and B are known matrices of appropriate sizes. Then, the corresponding discrete-time version of equation (1) is given as [4] xk + 1 = Fxk + G0k ual, k + G1k ual, k1 ð4Þ Ð T ttot xk = x(kT ), F = eAT , G0k = 0 k eAs dsB, G1k = Ðwhere T a eAs dsB, ttot k , and ul, k denote the S/A delay and the T ttot k actuator input, respectively, in response to the sensor signal xk . If the packet dropout is also considered, ual, k can be expressed as   tot ual, k = utot k ul, k + 1  uk ul, k1

ð5Þ

where ul, k and utot k are the control signal of node l and the value of switch utot , respectively, at the kth step. Equation (5) indicates that the actuator remains to use the latest control signal ul, k1 when ul, k is not generated due to the S/C dropout or lost due to the C/A dropout; otherwise, the newly arrived control signal ul, k is used. If the consecutive packet dropout happens, iteratively using equation (5), the general formulation of ual, k is given as ! k 1 k Y X   tot a tot tot ul, k = uk ul, k + 1  ui ð6Þ uj ul, j j=0

i=j+1

which means that ual, k = ul, km when tot ul = 0, l = k, k  1,    , k  m + 1 and utot km = 1, that is, the latest available control signal ul, km is used when ul, k and the previous (m  1) control signals are all lost or not generated. Substituting equations (6) into (4), the discrete-time control process with delays and packet dropouts can be expressed as xk + 1 = Fxk +

k X j=0

where

bj, k ul, kj

ð7Þ

4

International Journal of Distributed Sensor Networks

bj, k =

8 > G0k utot > k , > >  0   > < G 1  utot + G1 utot , k k k k1

where the time-varying coefficient matrices Ck and Dk are given as

j = 0, j = 1,

!

> kq 1  tot P Q  > q > > G 1  utot ukj , > i k : q=0 i = kj + 1

2 others ð8Þ

which is determined by the S/A delay ttot k and packet , i = k, k  1,    , k  j. dropouts utot i Note that there are many works to model the discrete-time control process with packet dropouts. However, most of them ignore the network-induced delay or directly set the actuator input to be zero when the packet dropout happens, which is the typical way to simplify the analysis. Here, considering the eventdriven controller and actuator, the plant state loss and the control signal loss can result in no signal (i.e., no event) arriving at the controller and the actuator, respectively. In this case, the actuator input remains the same as the last control signal, but cannot be directly set to zero. In addition, the effect of network-induced delays is also considered. Thus, we formulate the discrete-time control process as in equation (7). In this article, we focus on the optimal control problem of the WSAN with stochastic network-induced delays and packet dropouts. Using a quadratic cost function as in Robinson and Kumar,14 Xin et al.,18 and Quevedo et al.,19 the design of the optimal control is to minimize the cost function, that is "

N 1   X min JN = E xTN QN xN + xTk Qxk + uTl, k Rul, k

ful, k g

F

b1, k

b2, k



bk, k

0 0

0 IK

0 0

 

0 0

0 .. . 0

0 .. . 0

IK .. . 0

 .. . 

0 .. . IK

6 6 6 6 6 Ck = 6 6 6 6 4

3

3 2 b0, k 7 7 7 6 7 6 IK 7 7 7 6 7 7 6 7, Dk = 6 0 7 7 6 . 7 7 6 . 7 7 4 . 5 5 0 ð12Þ

where 0 is the zero matrix with appropriate size, and IK is the K 3 K identity matrix. Note that Ck and Dk are independent of the location of the controller. Then, the optimization problem in equation (9) can be rewritten as "  N zl, N + min JN = E zTl, N Q

ful, k g

N 1  X

 l, k + uT Rul, k zTl, k Qz l, k



# ,

k=0

s:t: zl, k + 1 = Ck zl, k + Dk ul, k ð13Þ

where 0

QN B 0 N = B . Q @ ..

0 0 .. .

0

0

#

0 1 Q  0 B0  0C  =B . . C, Q .. @ .. . .. A 0  0

0 0 .. . 0

1  0  0C .C .. . .. A  0 ð14Þ

k =0

ð9Þ

where E is the expectation operator over the distributions of all delays and packet dropouts, N is the total number of sampling instants, QN †0 is a symmetric positive semi-definite weight matrix, and Q  0 and R  0 are symmetric positive definite weight matrices.

Lemma 1. Assume that function lðx, y, uÞ has a unique minimum with respect to u 2 U for all x 2 X and y 2 Y .22 Then, we have min E½l(x, y, u) = E minu fl(x, y, u)g

ð15Þ

u(x, y)

Define

Optimal control law design

VL =

In this section, the optimal control law is derived for the WSAN in equation (9) subject to the constraint equation (7).

min

ul, k , k = L, , N 1

" E

 N zl, N zTl, N Q

+

N 1  X

 l, k zTl, k Qz

+ uTl, k Rul, k



# ð16Þ

k =L

We next derive the expressions for the optimal control law.

Optimal control law derivation Define  zl, k = xTk

uTl, k1

uTl, k2

   uTl, 0

T

2 RM + kK ð10Þ

L = N. When L = N, we have h i VN = E zTl, N SN zl, N

Then, equation (7) can be rewritten as zl, k + 1 = Ck zl, k + Dk ul, k

ð11Þ

N where SN = Q

ð17Þ

Wang et al.

5

L = N – 1. When L = N  1, based on Lemma 1, we get VN 1 = min ul, N 1  h  i   l, N 1 + uT + V E zTl, N 1 Qz Ru z l, N1 N l, N 1 l, N 1 n  T  l, N 1 + uT = min E zl, N1 Qz Rul, N 1 l, N 1 ð18Þ ul, N1  T   + E W SN ðW Þ zl, N 1 "

# 

 zl, N 1  zl, N 1 T = E minul, N 1 H zl, N1 ul, N 1 ul, N 1 

where

k

k

k

ð26Þ

Hereh Sk is symmetric, and the corresponding value of i T Vk = E zl, k Sk zl, k . From equations (25) and (26), we observe that the optimal control law Lk is obtained by backward recursion, and the optimal control strategy is linear with the plant state and all past control signals.

Some discussions

W = ðCN1 zl, N 1 + DN 1 ul, N 1 Þ,     !  + E C T SN CN 1 E CNT 1 SNT DN 1 Q N 1     H= R + E DTN 1 SN DN 1 E DTN 1 SN CN 1 ð19Þ

Then, the optimal control law minimizing equation (18) with respect to ul, N 1 is given as [23] ul, N 1 = LN1 zl, N1

ð20Þ

where   1  T  LN1 = R + E DTN 1 SN DN1 E DN1 SN CN 1 ð21Þ

At the same time, substituting ul, N 1 in equation (20) into equation (18), we have h i VN 1 = E zTl, N 1 SN 1 zl, N 1

ð22Þ

   + E C T SN CN1 SN 1 = Q N 1    LTN 1 R + E DTN 1 SN DN 1 LN 1

ð23Þ

where

which is symmetric, that is,SNT 1 = SN 1 L = N  2, :::, 1, 0 . When L = N  2, from equations (16) and (22), we have VN 2 = min ul, N 2  h  i  T  l, N2 + uT E zl, N 2 Qz Ru + V , z l, N 2 N 1 l, N2 l, N2 ð24Þ

We can see that VN 2 has the same form as equation (18). Repeating the derivation from equations (18) to (23), the optimal control law of ul, N 2 can be obtained. And similarly for other steps of ul, k , 8k. We then have ul, k = Lk zl, k , k = 0, 1,    , N  1

where

  1  T  Lk = R + E DTk Sk + 1 Dk E D k Sk + 1 C k ,       + E C T Sk + 1 Ck  LT R + E DT Sk + 1 Dk Lk Sk = Q

ð25Þ

In this subsection, some issues are discussed for design of the optimal controller. Optimal location for the controller. The design of the optimal controller is obtained in equation (25) and we observe that Lk is independent of the controller location l, since all parameters such as Ck and Dk are independent of l. In other words, the optimal control law is the same for all possible placements of the controller. Then, as the full plant state information is needed for the optimal control law, the optimal location should be collocated with the sensor, because the sensor is able to obtain the perfect plant state information, while only partial state information is available for other nodes due to the S/C delays and packet dropouts. Note that some articles assume that the actuator also has the perfect state information directly obtained from the plant. In this case, either the sensor or the actuator can be chosen as the optimal location for the placement of the controller. State noise. The disturbance usually exists at the plant. If the state noise is considered, equation (4) can be rewritten as xk + 1 = Fxk + G0k ual, k + G1k ual, k1 + v

ð27Þ

where v is the white noise with zero mean and covariance matrix sv . Using equations (27) instead of (4) as the dynamic process for design of the optimal controller, similar to that in subsection ‘‘Optimal control law derivation,’’ we can derive the same optimal control law as equation (25), and the value of Vk is changed to be h i Vk = E zTl, k Sk zl, k + ak

ð28Þ

  where ak = tr Sk11+ 1 sv + ak + 1 , aN = 0, and Skij denote the (i, j)th block of Sk . The cost function in   equation (9) can be derived as JN = V0 = 1E xT0 S011 x0 + a0 , where x0 is the initial

6

International Journal of Distributed Sensor Networks

Figure 3. Signal timings.

value of the plant state. Thus, the state noise introduces an extra cost a0 , but has no influence on the design of the optimal controller. Suboptimal solution. From equations (25) and (26), a drawback with the optimal control law is the complicated matrix Lk , since the dimension of zl, k increases when k becomes larger. In order to reduce the computational complexity, one approach is to truncate the effect of packet dropouts that only keep the ones with most contributions. Then, equation (6) is modified to be ual, k

= utot k ul, k

k1 X

+

j = kW

! k Y   tot tot 1  ui uj ul, j ð29Þ i=j+1

which means that at most the effect of W consecutive packet dropouts is considered. Using equation (29), it is equivalent to set ul, j [0 when j\k  W . In this case, the optimal control law in equation (25) can be simplified as ð30Þ

ul, k = Lk zl, k

where  T zl, k = xTk uTl, k1 uTl, k2    uTl, kW 2 RM + WK ,   1  T  Lk = R + E DTk Sk + 1 Dk E Dk Sk + 1 Ck ð31Þ

and the related parameters Dk and Sk have the same expressions defined in equations (12) and (26), which are reduced to be ðW + 1Þ 3 1-block and ðW + 1Þ 3 ðW + 1Þ-block matrices, respectively, and the other parameter Ck is simplified to be ðW + 1Þ 3 ðW + 1Þ-block matrix as follows 2

F 60 6 60 6 Ck = 6 0 6 6 .. 4 . 0

b1, k 0 IK 0 .. .

b2, k 0 0 IK .. .

    .. .

bW 1, k 0 0 0 .. .

0

0



IK

3 bW , k 0 7 7 0 7 7 : 0 7 7 .. 7 . 5 0

ð32Þ

From the definition of zl, k in equation (31), its dimension RM + WK is independent of k. Thus, the drawback that the computational complexity is determined by the value of k is solved.

Extension to the long delay case In this section, we extend the result with short networkinduced delay in section ‘‘Optimal control law design’’ to the case when the delay is larger than one sampling period. Here, we assume that the S/A network-induced delay is bounded that ttot \hT, where h is a positive integer. Then, during one sampling interval, the number of received control signals on the actuator is between 0 and h. The signal timing is illustrated in Figure 3. to show that the actuation time set is ftkh2 , tkh1 , :::, tk , tk1 g. Note that the actuation time tj , j = k  h  2, k  h  1, :::, k only depends on the beginning of the sampling jT and the corresponding S/A delay ttot j . Thus, tj is independent of the location of the controller. In addition, the long delay results in a disorder in the control signal arrivals at the actuator, for example, the control signal ul, k arrives before ul, k1 . In this case, we will ignore the control signal ul, k1 and still use the newer control signal ul, k even after ul, k1 arrives at the actuator. Then, for a given sampling interval ½kT , (k + 1)T Þ, suppose that there are mk effective control signals from controller i, where 0  mk  h, which arrive at time instants th1 , th2 , :::, thmk , where k  h\h0 \h1 \h2 \    \hmk \k + 1, and h0 denotes the index of the active control signal at time kT (see Figure 4 for an example). Then, the control signals will be actuated in the order of ul, h0 , ul, h1 , :::, ul, hml, k , and   the corresponding actuation periods are kT , th1 , h    th1 , th2 ,    , thmk , ðk + 1ÞT . Thus, integrating equation (3) over a sampling interval ½kT , (k + 1)T Þ, we obtain the following discrete-time linear system parameterized by the network-induced delays xk + 1 = Fxk +

h X j=0

Gjk ual, kj

ð33Þ

Wang et al.

7

Figure 4. An example of arrivals in ½kT, (k + 1)T Þ. In this case, h0 = k  3, h1 = k  1, h2 = k. As uk3 is active in the beginning, uk2 arrives later than uk (and uk1 ), it is ignored. The arrival instants of effect control signals are tk1 and tk .

 j, k , in equation (25). In addition, it can be seen that, b  same as bj, k , is independent of the location of the controller. Thus, for the case of long stochastic delay, similar properties for the short delay case also hold: The optimal control law is the same for all controller placements, and the optimal location should be at the sensor.

Simulation results

Figure 5. WSAN-based LFC system for power grid.

where

Gjk

=

8 > > > > > > >
> Ð ðk + 1ÞT Aððk + 1ÞT sÞ > > > e dsB, > > thmk : 0,

j = k  h1 , .. . j = k  hmk ,

ð34Þ

otherwise

Note that Gjk is random due to the stochastic S/A delays. Based on equations (6) and (33), the discrete-time control process with long delays is given as xk + 1 = Fxk +

k X

 u b j, k l, kj

ð35Þ

j=0

where  j, k = b 8 > < j tot > : rj Gk ukj +

h P q=0

G0k utot k , kq Q q

Gk

i = kj + 1



1  utot i



! utot kj ,

j = 0, others ð36Þ

and rj = 1 when j = 1, 2, :::, h, and otherwise rj = 0. We observe that equations (7) and (35) have the  j, k instead of bj, k , the same same form. Then, using b expression of the optimal control law can be derived as

In this section, we provide simulation studies on the application of the proposed optimal controller scheme with stochastic network-induced delays and packet dropouts to the LFC system in power grid.24,25 The typical LFC system is composed of speed governor, turbine, generator, and LFC controller (included in the wireless sensor and actuator network), which is illustrated in Figure 5. The objective is to optimally adjust LFC controller u to keep the frequency deviation Df within prescribed limits. The deviations of the generator-turbine-governor system can be represented by two time constants, Tt of the turbine and Tg of the governor. The generator response is considered to be instantaneous in comparison with the time constants of turbine and governor, which can be written as d 1 1 DPg = DPg + DXg , dt Tt Tt d 1 1 DXg = DXg + DPc dt Tg Tg

ð37Þ

where DPg , DXg , and DPc are the deviations of generator mechanical output, valve position, and generator output, respectively. The deviation of frequency Df is given as Kp d 1 Df = Df + DPg dt Tp Tp

ð38Þ

where Kp is the electric system gain and Tp is the electric system time constant. The system state vector can be defined as x(t) = ½ DPc

Df

DPg

DXg T

ð39Þ

The sensor is required to measure the fourdimensional corresponding system state, and then send

8

International Journal of Distributed Sensor Networks In the simulation, a 4-hop wireless network composed of the sensor, actuator, and three relay nodes is chosen for simulation. For each hop, it is assumed that the packet dropout follows the Bernoulli distribution with p = 0:9, and the network-induced delay is uniform in ½0, aT , 0  a  0:5 where the maximum S/A delay is 2T . In addition, we set the initial value of the plant state x0 = ½ 0:25 0:15 0:2 0:1 T , Kp = 1, Tp = 0:2, Tt = 0:3, Tg = 0:08, and 2

1 60 QN = Q = 6 40 0

Figure 6. Performance comparison for various controller locations.

Figure 7. Performance comparison of the proposed scheme with existing approaches.

to the LFC controller. Once the state signal has arrived at the controller, a new control signal is calculated and sent to the actuator node. Then, the linear dynamic control model can be described as x_ (t) = Ax(t) + Bua (t  t)

ð40Þ

where ua is the actuator input corresponding to the LFC control signal u, which is affected by the packet losses introduced by the WSAN as shown in equation (6). Based on equations (37) and (38) and Figure 5, we obtain the system parameters B = ½ 1 1 0 0  and 2

0 6 0 A=6 4 0 1=Tg

0 1=Tp 0 0

0 Kp =Tp 1=Tt 0

3 0 0 7 7 1=Tt 5 1=Tg

ð41Þ

0 1 0 0

0 0 1 0

3 0 07 7, R = 1 05 1

ð42Þ

and the sampling period and duration are chosen as T = 0.05 and N = 50, respectively. Figure 6 shows the performance comparison when the controller is located at different nodes such as the sensor (the first node) and relay nodes (the second, third, and fourth nodes). We observe that the cost can be reduced when the controller collocates with the sensor, which indicates that the imperfect plant state information caused by the network-induced delays and packet dropouts introduces performance degradation, and the sensor is the optimal controller placement. In addition, the system performance is degraded significantly with the network-induced delay increasing. This reveals that the effect of delays should be considered in the design of the optimal control law. Figure 7 shows the performance comparison for three cases: our proposed scheme with the optimal controller location, the approach in Robinson and Kumar14 where the controller collocates with the actuator, and the LQ optimal control strategy proposed in Garone et al.16 ignoring the network-introduced delay. We can observe that our proposed scheme with the optimal controller location outperforms the other two schemes in the sense that it has a lower total cost. It can also be seen that the cost of the approach proposed in Garone et al.16 without considering the effect of network-induced delay increases more rapidly than the others. This is because this scheme cannot effectively deal with the network-induced delay so that both network-induced delays and packet losses should be investigated. The performance of the suboptimal approach is shown in Figure 8, where the controller is assumed to be collocated with the sensor. It can be seen that when we truncate the effect of packet dropouts, the performance degradation is significantly decreased with larger W , and the cost is close to the optimal one even when the value of W is small (W = 4). Thus, keeping a proper length of the consecutive packet dropout is also an effective approach to reducing the computational

Wang et al.

9 corresponding performances are investigated by the LFC system in power grid. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

Figure 8. Performance comparison for suboptimal control design.

Figure 9. The effect of plant state noise.

complexity and meanwhile maintaining the nearoptimal performance. In Figure 9, the effect of the plant state noise is investigated. Here the covariance matrix of the noise is set as sv = sIM , where s is chosen as 0.0005, 0.001, 0.0015, and 0.002. It can be seen that the noise introduces an extra cost to degrade the system performance, and the extra cost increases with the value of s, which well matches the discussion on the cost function in equation (28).

Conclusion We have considered the design of the optimal control for the WSAN in the presence of stochastic networkinduced delays and packet dropouts. In particular, the optimal control solution is derived as a linear function of the current plant state and all past control signals, and the optimal controller should be collocated with the sensor. Other issues such as the state noise and suboptimal solution are also discussed, and the

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by National Natural Science Foundation of China under Grant Nos 61701010, 61601330, and 61671029, and the Foundation of Beijing Municipal Commission of Education KM201610005004.

References 1. Chai S, Liu GP, Rees D, et al. Design and practical implementation of Internet-based predictive control of a servo system. IEEE T Contr Syst T 2008; 16(1): 158–168. 2. Bullo F, Cortes J and Martinez S. Distributed control of robotic networks: a mathematical approach to motion coordination algorithms. Princeton, NJ: Princeton University Press, 2009. 3. Zhang W, Branicky MS and Phillips SM. Stability of networked control systems. IEEE Contr Syst Mag 2001; 21(1): 84–99. 4. Nilsson J, Bernhardsson B and Wittenmark B. Stochastic analysis and control of real-time systems with random time delays. Automatica 1998; 34(1): 57–64. 5. Hu S and Zhu W. Stochastic optimal control and analysis of stability of networked control systems with long delay. Automatica 2003; 39(11): 1877–1884. 6. Hespanha J, Naghshtabrizi P and Xu Y. A survey of recent results in networked control systems. Automatica 2007; 95(1): 138–162. 7. Zhang L, Gao H and Kaynak O. Network-induced constraints in networked control systems1a survey. IEEE T Ind Electron 2013; 9(1): 403–416. 8. Sinopoli B, Schenato L, Franceschetti M, et al. Kalman filtering with intermittent observations. IEEE T Automat Contr 2004; 49(9): 1453–1464. 9. Quevedo D and Nesic D. Input-to-state stability of packetized predictive control over unreliable networks affected by packet-dropouts. IEEE T Automat Contr 2011; 56(2): 370–375. 10. Hadjicostis CN and Touri R. Feedback control utilizing packet dropping network links. In: Proceedings of the 41st IEEE conference on decision and control, Las Vegas, NV, 10–13 December 2002, pp.1205–1210. New York: IEEE. 11. Tabbara M, Nesic D and Teel A. Stability of wireless and wireline networked control systems. IEEE T Automat Contr 2007; 52(9): 1615–1630. 12. Hart Communication Foundation, http://www.hart comm.org/ 13. ISA 100 WIRELESS, http://www.isa100wci.org/ 14. Robinson CL and Kumar PR. Optimizing controller location in networked control systems with packet drops. IEEE J Sel Area Comm 2008; 26(4): 661–671.

10 15. Gupta V, Dana AF, Hespanha JP, et al. Data transmission over networks for estimation and control. IEEE T Automat Contr 2009; 54(8): 1807–1819. 16. Garone E, Sinopoli B, Goldsmith A, et al. LQG control for MIMO systems over multiple erasure channels with perfect acknowledgment. IEEE T Automat Contr 2012; 57(2): 450–456. 17. Lin H, Su H, Shi P, et al. LQG control for networked control systems over packet drop links without packet acknowledgment. J Frankl Inst 2015; 352(11): 5042–5060. 18. Xin K, Cao X, Chen J, et al. Optimal controller location in wireless networked control systems. Int J Robust Nonlin 2015; 25(2): 301–319. 19. Quevedo DE, Johansson KH, Ahlen A, et al. Adaptive controller placement for wireless sensor-actuator networks with erasure channels. Automatica 2013; 49(11): 3458–3466. 20. Wang Z, Wang X, Liu L, et al. Optimal state feedback control for wireless networked control systems with

International Journal of Distributed Sensor Networks

21.

22. 23.

24.

25.

decentralised controllers. IET Control Theory A 2015; 9(6): 852–862. Wang Z, Wang X and Liu L. Stochastic optimal linear control of wireless networked control systems with delays and packet losses. IET Control Theory A 2016; 10(7): 742–751. Astrom KJ. Introduction to stochastic control theory. New York: Academic Press, 1970. Astrom KJ and Wittenmark B. Computer-controlled systems theory and design. 3rd ed. Upper Saddle River, NJ: Prentice Hall, 1997. Dong L and Zhang Y. On design of a robust load frequency controller for interconnected power systems. In: Proceedings of the IEEE American control conference, Baltimore, MD, 30 June–2 July 2010, pp.1731–1736. New York: IEEE. Fosha CE and Elgerd OI. The megawatt-frequency control problem: a new approach via optimal control theory. IEEE T Power Ap Syst 1970; PAS-89(4): 563–577.