International Journal of Power Electronics and Drive Systems (IJPEDS)

Vol. 13, No. 4, December 2022, pp. 2086~2097

ISSN: 2088-8694, DOI: 10.11591/ijpeds.v13.i4.pp2086-2097  2086

Efficient detection of faults and false data injection attacks in

smart grid using a reconfigurable Kalman filter

Prakyath Dayananda1, Mallikarjunaswamy Srikantaswamy2, Sharmila Nagaraju3, Rekha Velluri4,

Doddananjedevaru Mahesh Kumar5
1
Department of Electrical and Electronics Engineering, SJB Institute of Technology, Bangalore, India
2
Department of Electronics and Communication Engineering, JSS Academy of Technical Education, Bangalore, India
3
Department of Electrical and Electronics Engineering, Sri Jayachamarajendra College of Engineering, Mysore, India
4
Department of Computer Science and Engineering, CHRIST (Deemed to be University), Bengaluru, India
5
Department of Electronics and Instrumentation Engineering, JSS Academy of Technical Education, Bangalore, India

Article Info ABSTRACT

Article history: The distribution denial of service (DDoS) attack, fault data injection attack
(FDIA) and random attack is reduced. The monitoring and security of smart
Received Mar 17, 2022 grid systems are improved using reconfigurable Kalman filter. Methods: A
Revised Aug 28, 2022 sinusoidal voltage signal with random Gaussian noise is applied to the
Accepted Sep 19, 2022 Reconfigurable Euclidean detector (RED) evaluator. The MATLAB
function randn() has been used to produce sequence distribution channel
noise with mean value zero to analysed the amplitude variation with respect
Keywords: to evolution state variable. The detector noise rate is analysed with respect to
threshold. The detection rate of various attacks such as DDOS, Random and
False data injection attack false data injection attacks is also analysed. The proposed mathematical
Kalman filter model is effectively reconstructed to frame the original sinusoidal signal
Random attack from the evaluator state variable using reconfigurable Euclidean detectors.
Reconfigurable Euclidean
detector This is an open access article under the CC BY-SA license.
Smart grid

Corresponding Author:
Mallikarjunaswamy Srikantaswamy
Department of Electronics and Communication Engineering, JSS Academy of Technical Education
JSSATE-B Campus, Dr. Vishnuvardhan Road, Uttarahalli - Kengeri Main Road
Srinivaspura-Post Bengaluru – 560060 Karnataka, India
Email: [email protected]

1. INTRODUCTION
The Power grid is considered to be a significant backbone of infrastructure, which has a profound
effect on the economy and the day-to-day routines. Fiascos in the power grid normally have shattering
effects. With the beginning of fresh skills, the out-of-the-way power grid is updated by a grid that is a
distinctive smart cyber-physical system (CPS) that includes additional implanted smartness and networking
competence. Sensors are furnished all over the system to observe different grid features, like the meter &
voltage fluxes in such arrangements [1]−[3]. The gathered data by the sensors aids to give a reaction to the
physical power grids. So, that kind of a CPS comprises two-approach messages among the controller scheme
and the physical apparatuses as depicted in Figure 1. Numerous evolving attacks precisely aiming at the
control and communication arrangements in smart grid are uncovered. A broad approach to detect physical
altering is done by a process of installing an evaluator along with a corresponding detector in the given
controller. A remarkable variance among the estimated and measured states indicates a likely attack on the
structure. Here, we showcase a framework of security utilizing the Kalman filter (KF) for a given smart grid.
The KF produces assessments for state variables by means of the mathematical prototype for the power grid
& the information got by the system of sensors is installed so as to observe the power grid. Then we can also

Journal homepage: http://ijpeds.iaescore.com

Int J Pow Elec & Dri Syst ISSN: 2088-8694  2087

use a χ2-detector which can further be used to identify the inconsistencies amid the assessed data & the
experimental data and trigger alarms [4].
Nevertheless, the learning depicts that the χ2-detector can’t identify the statistically resultant false
data injection attack. We broadly examine this attack, along with the planned KF framework and project a
supplementary detection method by means of the Euclidean distance metric [5], [6]. The main objectives
include i) we plan a mathematical prototype along with the KF to identify likely attacks & errors on the
system of smart grid, ii) then examine the functioning of the investigative technique χ2-detector, in
recognizing errors & arbitrary attacks, and iii) we evaluate the restraint of the χ2 detector in sensing the
analytically resultant false data injection attack and consequently project a fresh detector of Euclidean to be
joined with KF and d) then showcase the efficacy of the planned methods through widespread simulations &
study on real-world systems. The remaining work is structured as below. Section 2 shows the motivation &
the associated work on smart-grid security. Section 3 shows the planned structure, the measured prototype of
the power grid & the KF estimator. Section 4 shows the two detectors employed in the structure so as to
identify different attacks in the arrangement. In section 5, outcomes of the planned framework and the
interpretations are detailed. Lastly, section 4 details the conclusion.

Figure 1. Fundamental block diagram of a smart grid system

2. MOTIVATION AND RELATED WORK

In this segment, we examine different security aspects deliberated in the literature. The modern
studies on smart grid security could be largely classified into three sets. The research in the first group pacts
with the wired or wireless security of networking between cyber constituents in the smart grid. The work
inside second group would examine the early detection of abnormalities in the structure. The early anomaly
detection methods can forehandedly safeguard the system. The research in the third group smears the control
theories in the security procedure utilizing different state assessment & revealing methods.
A signature-dependent message validation system was projected, that works in the multicast
authentication format in order to lessen the size of signature & bandwidth of communiqué at the price of
augmented calculation. The projected method includes detecting, reacting, data recollecting & alarm
supervision mechanisms. An error inside the smart grid scheme is constantly shown in the method of
alteration in voltage, phase or current. Prevailing security methods are either i) not feasible, ii) mismatched
with the smart grid, iii) not suitably scalable, or iv) not sufficient. Our research shows a structure, dependent
on a state-space system obtained by the voltage stream reckonings, to secure various kinds of attacks &
errors, corresponding to the Injection attack of the false data. We project an altered detector dependent on the
metric of Euclidean distance to identify a complex Injection attack of the given false data over the power grid
mechanism.

3. PROJECTED STRUCTURE FOR SMART GRID BY MEANS OF RED

In this segment, we showcase the complete report of the structure of the given security for the smart
grid which is utilizing the KF. The structure is proficient in identifying different assaults, including short &
long-term arbitrary attacks including the development of a state-space prototype by the three-phase
Efficient detection of faults and false data injection attacks in smart grid … (Prakyath Dayananda)
2088  ISSN: 2088-8694

sinusoidal voltage reckonings [7]−[9]. Figure 2 depicts the projected structure of security, where we can see
that KF assesses the figures for the given state parameters depending on state of the system and the statistics
from various values of sensors. The KF produced projected values and the detected figures alongside
variables of state are then given inside the detector. After this, the two-state vectors are equated by the
detector. If the two vary from each other considerably and are above an assured pre-calculated threshold, an
alarm to imply a likely smart grid attack is initiated by the detector.

Figure 2. Security execution protocols of the smart grid system

3.1. Prototype of state space

The power structure installs sensors, like phasor units of measurement, so as to assess the system
state at different places and stint to make sure of the even operation of the power scheme. the sinusoidal
voltage mathematical model is given in (1). the three phases voltage signals are given in (2) and (3)
respectively [10], [11].

𝑆1 (𝑡) = 𝐴𝑓 cos(𝑤𝑡 + 𝜑) (1)

2𝜋
𝑆2 (𝑡) = 𝐴𝑓 cos(𝑤𝑡 + 𝜑 − ) (2)
3

4𝜋
𝑆2 (𝑡) = 𝐴𝑓 cos(𝑤𝑡 + 𝜑 − ) (3)
3

Extension of (1) as shown in (4).

𝑆1 (𝑡) = 𝐴𝑓 ∗ 𝑐𝑜𝑠𝑤𝑡 ∗ 𝑐𝑜𝑠𝜑 − 𝐴𝑓 ∗ 𝑠𝑖𝑛𝑤𝑡 ∗ 𝑠𝑖𝑛 𝜑 (4)

Where 𝐴𝑓 is described the amplitude function, 𝑤𝑡 is represented as the angular frequency and 𝜑 is identified
as phase angle with respect to time. When the angular frequency is constant with respect to the time then
amplitude and phase can be represented in state-space model is given in (5).

𝑆1 (𝑡) = 𝑔1 ∗ cos 𝑤𝑡 − 𝑔2 ∗ sin 𝑤𝑡 (5)

Where 𝑔1 = 𝐴𝑓8 cos 𝜑 and 𝑔2 = 𝐴𝑓𝑜 sin 𝜑 is the state variables at no delay condition in the model. The tiny
noise is applied to the system and this condition is given in (6).
𝑔1 (𝑡 + 1) 1 0 𝑔1 (𝑡)
[ ]=[ ][ ] + 𝑤(𝑡) (6)
𝑔2 (𝑡 + 2) 0 1 𝑔2 (𝑡)

In (6) can be represented in simplest form and it is given in (7).

1 0
𝑔(𝑡 + 1) = [ ] 𝑔(𝑡) + 𝑤(𝑡) (7)
0 1
𝑔1 (𝑡)
Where 𝑔(𝑡) = [ ] and 𝑤(𝑡) is described as process noise. At nonstationary deterministic condition the
𝑔2 (𝑡)
actual voltage signal is given in (8). Where ℎ(𝑡) is describes the actual voltage signal with respect to time
and 𝛤(𝑡) is represents the measurement noise.

Int J Pow Elec & Dri Syst, Vol. 13, No. 4, December 2022: 2086-2097
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2089

𝑔1 (𝑡)
ℎ(𝑡) = [𝑐𝑜𝑠𝑤𝑡 − 𝑠𝑖𝑛𝑤𝑡] [ ] + 𝛤(𝑡) (8)
𝑔2 (𝑡)

3.2. Reconfigurable Kalman filter (RKF)

Figure 3 depicts the control system alongside the KF entrenched on the approximation of the vector
of state & detector in order to do the identification of errors [12], [13]. The (9) represents the KF technique
1 0
where = [ ] , from in (8) it can be represented in simplest form and it is given in (10).
0 1
𝑔(𝑡 + 1) = 𝐴𝑔(𝑡) + 𝑤(𝑡) (9)

ℎ(𝑡) = 𝑉𝑐 (𝑡)𝑥(𝑡) + 𝑠(𝑡) (10)

Where ℎ(𝑡) is identified as sensor measurement vector, 𝑉𝑐 (𝑡) is represents the [𝑐𝑜𝑠𝑤𝑡 − 𝑠𝑖𝑛𝑤𝑡] and 𝑠(𝑡) is
described as a white gaussian noise at mean is zero and standard deviation ‘ρ’ it is not depend on process
noise and initial condition [14], [15]. The KF mean and covariance of the evaluator is defined by (11)-(14).

𝑠̂ (𝑡|𝑡) = 𝐸𝑠𝑡 [𝑔(𝑡), ℎ(𝑡), … … . ℎ(𝑡)] (11)

𝑠̂ (𝑡|𝑡 − 1) = 𝐸𝑠𝑡 [𝑔(𝑡), ℎ(𝑡), … … . ℎ(𝑡 − 1)] (12)

𝑃𝑡 (𝑡|𝑡) = ∑(𝑡|𝑡 − 1) (13)

𝑃𝑡 (𝑡|𝑡 − 1) = ∑(𝑡|𝑡 − 1) (14)

Where, 𝑠̂ (𝑡|𝑡) is represents the signal evaluator with respect to ‘t’, 𝑠̂ (𝑡|𝑡 − 1) is describes the signal
evaluator with respect to time ‘t-1’, 𝑃𝑡 (𝑡|𝑡) is identified as covariance of the Evaluator with respect to the ‘t’
and 𝑃𝑡 (𝑡|𝑡 − 1) is represents the covariance of the Evaluator with respect to the ‘t-1’. The KF Iteration
process is represented by (15) and (16).

𝑠̂ (𝑡 + 1|𝑡) = 𝐴𝑠̂ (𝑡) (15)

𝑃𝑡 (𝑡|𝑡 − 1) = 𝐴𝑃𝑡 (𝑡 − 1)𝐴𝑇 + 𝑍 (16)

Where 𝑠̂ (𝑡 + 1|𝑡) is represents the state and covariance of the evaluator with respect to t to t+1-time steps,
𝑃𝑡 (𝑡|𝑡 − 1) is describes the covariance of the evaluator with respect to t-1 to t and ‘Z’ is represents the
covariance matrix process noise [16], [17]. The RKF measuring updates are represented by (17)-(19).

𝐾𝐴 (𝑡) = 𝑃𝑡 (𝑡|𝑡 − 1)𝑉𝑐 (𝑡)𝑇 (𝑉𝑐 (𝑡)𝑃𝑡 (𝑡|𝑡 − 1)𝑉𝑐 (𝑡)𝑇 + 𝑅)−1 ) (17)

𝑃𝑡 (𝑡|𝑡) = 𝑃𝑡 (𝑡|𝑡 − 1) − 𝐾𝐴 (𝑡)𝑉𝑐 (𝑡)𝑃(𝑡|𝑡 − 1) (18)

𝑠̂ (𝑡) = 𝑠̂ (𝑡|𝑡 − 1) + 𝐾𝐴 (𝑡)(ℎ(𝑡) − 𝑉𝑐 (𝑡)𝑠̂ (𝑡|𝑡 − 1)) (19)

Where 𝐾𝐴 (𝑡) is described the reconfigurable Kalman gain and R represents the covariance matrix noise
analysis. The Kalman Gain before the evaluation is represented by (20), (21) and enhancement of (19) is
given in (22). The evaluation error 𝛿 (𝑡) is represented in (23).

𝑃 ≜ lim 𝑃𝑡 (𝑡|𝑡 − 1) (20)

𝑘→∞

𝐾𝐴 = 𝑃𝑡 𝑉𝑐𝑇 (𝑉𝑐 𝑃𝑡 − 1)−1 (21)

𝑠̂ (𝑡 + 1) = 𝐴𝑠̂ (𝑡|𝑡) + 𝐾𝐴 [(ℎ(𝑡 + 1) − 𝑉𝑐 𝐴𝑠̂ (𝑡) + 𝛽𝑢(𝑡)] (22)

𝛿 (𝑡) ≜ 𝑠̂ (𝑡) − 𝑠(𝑡) (23)

3.3. Model generalization

The state-space model is detailed in section 3. It could be widespread for power grid dimensions.
The voltage monitored at every bus could stay in the method of a sinusoidal [18], [19]. Let us study the three-
phase bus structure as shown in Figure 4.

Efficient detection of faults and false data injection attacks in smart grid … (Prakyath Dayananda)
2090  ISSN: 2088-8694

𝜁𝑥 = ∑𝑛𝑖=1|𝑆𝑥 ||𝑆𝑖 |𝑍𝑥𝑖 sin(𝜑𝑥 − 𝜑𝑖 ) − cos (𝜑𝑥 − 𝜑𝑖 ) (24)

𝜌𝑥 = ∑𝑛𝑖=1|𝑆𝑥 ||𝑆𝑖 |𝑍𝑥𝑖 sin(𝜑𝑥 − 𝜑𝑖 ) − cos (𝜑𝑥 − 𝜑𝑖 ) (25)

Where |𝑆𝑥 | is represented the voltage amplitude, |𝑆𝑖 | has described the phase, 𝑍𝑥𝑖 is identified as gain, 𝜁𝑥 is
represented the active power, 𝜌𝑥 has described reactive power and x is the number of system buses [20], [21].
To determine unknown variables in each system buses by (24) and (25).

Figure 3. Proposed power grid

Figure 4. Fundamental three bus structure

3.4. Model of attack

The model attack occurs when FDIA gets introduced to the smart grid system. It is able to control a
sub set of the sensor readings in the system. It is presumed that the invader is capable of controlling a
subdivision of the sensor evaluations in the structure. there are three types of attacks namely: i) DDoS,
ii) random, and iii) false data injection [22], [23].

3.5. DDoS attack

The DDoS attack is jamming the communication channel, compromising devices and flooding
packets in networks to avoid data transfer. This kind of assault is such that wherein an opponent extracts few
or every constituent of an unreachable control system. The bout of DDoS could be on control, sensor, or on
both data.

3.6. Random attack

Here, the assaults aren’t constructed to bypass the discovery procedure executed by the central
system. Such arbitrary attacks can be produced at any point in time.

ℎ′ (𝑡) = 𝑉𝑐 (𝑡)𝑔′ (𝑡) + 𝑠(𝑡) + ℎ𝑎 (𝑡) (26)

Where ℎ𝑎 (𝑡) is represented the random attack vector, ℎ′ (𝑡) is described as model observation and 𝑔′ (𝑡) is
identified as system process states [24].

3.7. False data-injection attack

It is alleged so as to know that the attacker is aware of the model of a given system, having variables
𝜌 R, A, 𝛽, 𝑉𝑐 and gain 𝐾𝐴 . The attacker could as well regulate a subdivision of sensors (Sbad). Where 𝜏 is
represented the sensor selection matrix 𝜏 = 𝑑𝑖𝑎𝑔(𝛾1 + 𝛾1 + 𝛾2 + 𝛾3 + 𝛾4 +, … … … . +𝛾𝑚 ) and 𝑥 ∈ 𝑆𝑏𝑎𝑑 .

Int J Pow Elec & Dri Syst, Vol. 13, No. 4, December 2022: 2086-2097
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2091

ℎ′ (𝑡) = 𝑉𝑐 (𝑡)𝑔′ (𝑡) + 𝑠(𝑡) + 𝜏ℎ𝑎 (𝑡) (27)

4. ATTACK DETECTOR
The RKF predictor computes the system state by means of the reckonings detailed in section 3.2. As
the readings of a given meter are evident for that state, the planned assessments and the authentic meter
evaluations are paralleled by the detector. In case the variance among the two is over an earlier calculated
threshold, an alarm is generated to inform a likely attack [25], [26].

4.1. 𝝌𝟐 -detector
The 𝜒 2 -detector is a traditional one which castoff with RKF. The 𝜒 2 --detector constructs 𝜒 2 -test
measurements from the RKF and parallels those with the threshold got from the customary 𝜒 2 -Table 1 [27].
Let the residue 𝑅(𝑡 + 1) at k+1 sec be determined by (28) and a simplified (30) is presented by (29). The
scalar test statistics of 𝜒 2 -detector is given in (30).

𝑅(𝑡 + 1) ≜ ℎ(𝑡 + 1) − ℎ̂(𝑡 + 1|𝑡) (28)

𝑅(𝑡 + 1) ≜ ℎ(𝑡 + 1) − 𝑉𝑐 (𝐴𝑠̂ (𝑡)) (29)

𝑤(𝑡) = 𝑅(𝑡)𝑇𝐵 (𝑡)𝑅(𝑡)

Where 𝑤(𝑡) is represented as the precomputed threshold and B (t) is described as the covariance matric of R
(t). The reconstructed sinusoidal signals from evaluator of the reconfigurable Euclidean detector. The
comparison analysis has been done with conventional methods by (30). Where 𝜁 is represented the amplitude
and 𝜌 is described the evaluated voltage signal amplitude. Table 1 shows the experimental setup of 𝜒 2 -
detector used PKF.

𝑑(𝜁, 𝜌) = √(𝜁1 − 𝜌1 )2 + (𝜁1 − 𝜌1 )2 + (𝜁1 − 𝜌1 )3 + ⋯ … … … … … + (𝜁𝑛 − 𝜌𝑛 )2 (30)

Table 1. Reconfigurable Kalman filter experimental setup

Particular Quantity
Initial covariance matric 𝜁 (0|0) Identified matrix
Frequency 65 Hz
The initial value for 𝑠1 (0) 0
The initial value for 𝑠2 (0) 0
Amplitude 1 Volt
Sampling frequencies 2.5 Hz

4.2. Detector executing the distance metric of Euclidean

The false data injection assault is sensibly made to avoid the numerical detector, like the 𝜒 2 -
detectors. So, to identify such kinds of assaults, we acclaim a reconfigurable euclidean-based detector, that
computes the aberration of the experiential figures compared to the assessed figures. To implement the
reconfigurable euclidean detector, initially, sinusoidal signals are built from the state assessments and then
equated with the quantities got from the sensors as depicted. If the variance among the two is more than the
threshold ‘3α’ where α is represented the standard deviation, as in the situation of the 𝜒 2 -detector, an alarm
is produced. To reduce to 99.85% of false positives obtained because of noise, we fix the threshold.

5. IMPLEMENTATION AND EVALUATION OF PERFORMANCE

We executed the RKF Evaluator, Euclidean detector, and 𝜒 2 -detector making use of MATLAB. The
research setup and the preliminary figures are depicted in Table 1. A 65Hz signal of the sinusoidal voltage
having arbitrary Gaussian noise is produced and given to the RKF estimator by way of the input. The input &
the consequent sinusoidal signal got utilizing the state assessments are shown in Figures 5 to 9.

5.1. Attack/error detection utilizing the 𝝌𝟐 −detector

Figure 5 depicts the simulation consequences utilizing the 𝜒 2 -detector in the lack of attacks for
some amount of duration. We can see that the assessed figures got from the KF estimator overlay with the
input signal showing there is no change amongst the projected and the experimental figures. The RKF
functions iteratively by amending its assessments utilizing the state-space model and the values got & the
Efficient detection of faults and false data injection attacks in smart grid … (Prakyath Dayananda)
2092  ISSN: 2088-8694

assessments slowly congregate through the input signal. In the time of assaults, the projected assessments
will not agree with the experiential analysis and w(t) surpasses the threshold as depicted in Figure 6 depicts a
short-duration attack being identified by the structure. Figure 7 depicts the discovery of the attack of the
DDoS.

Figure 5. No attack/fault signal transfer response using χ2 -detector

Figure 6. random attack for a short period transfer response using χ2 –detector

5.2. False data injection attack detection using RKF

Here, it so happens that it injects forged sensor measurements which can mislead the system by
executing the RKF estimator with the χ2-detector. The attack sequence can be obtained from in (31). Where
‘n’ is the represents the measurement of state space, ℎ∗ = 𝑉𝑐 𝑠, 𝑀 = 𝑚𝑎𝑥𝑖=, 1,2,3, 4…...n-1,

𝜆(𝑥+1)
ℎ𝑎 (𝑛 + 𝑡) = ℎ𝑎 (𝑡) − ℎ∗ (31)
𝑀

The source of the assault arrangement confirms that it overcomes the detector and upsurges the fault in the
assessment of the state. The second subgraph in Figure 7 depicts the behavior of the 𝜒 2 -detector beneath the
injection attack of the false data. We observe the approximations don’t match with the experimented figures
in the top subgraph in Figure 7 Nevertheless, w(t) never surpasses the threshold. We talk about this
disadvantage in the subsequent phase by utilizing the Euclidean detector that could detect such attacks by
continually observing the variation amongst the estimated and the experimented values.

Int J Pow Elec & Dri Syst, Vol. 13, No. 4, December 2022: 2086-2097
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2093

1.5 W(t)
8
Thershold
Input signal
1.0 Evaluated Signal

6
0.5

Simulated Signal
Simulation Signal

0.0 4

-0.5
2

-1.0
0

-1.5
0.00 0.02 0.04 0.06 0.08 0.10 0.00 0.02 0.04 0.06 0.08 0.10

Time (s) Time (A)

Figure 7. DDoS attack for a short period transfer response using χ2 –detector

5.3. False data injection attack discovery utilizing the Euclidean detector
This detector equates the alteration among the data experimented and the assessed data depending
on the metric of the Euclidean distance. Nevertheless, to evade fake alarms due to dimension faults, we set
the threshold to 3α as detailed in section IVB. Figure 8 shows the graph of the metric of the Euclidean
distance while an attack is not there in the structure and the below subgraph in Figure 8 shows the plot when
false data injection assault is there inside the structure.

Figure 8. No attack/fault signal transfer response using reconfigurable Euclidean distance

5.4. Load change

In the prototype obtained, it is presumed that the load in the system is persistent. If at all we have a
load change, then, there will be an alteration in the signal voltage through the buses. In case we know the
load profile, then the change in amplitude of voltage produced because of the load change can be predicted.
The factors inside the RKF can be attuned to reproduce the alteration inside the voltage because of the
alteration in load. It permits us to get assessments for the state variables subsequent to the change in load.
Figure 9 depicts that the assessments meticulously trail the signal along with the load alteration at time step
0.08, the random bout is identified by the 𝜒 2 detector & Euclidean detector in such situation.

Efficient detection of faults and false data injection attacks in smart grid … (Prakyath Dayananda)
2094  ISSN: 2088-8694

5.5. Χ2-Detector versus reconfigurable Euclidean detector

The likelihood of assault discovery in either of detectors is mainly reliant on the assessment of the
threshold. In 𝜒 2 - detector, the verge is got from the 𝜒 2 Table 1 Likewise, in reconfigurable Euclidean
detector, the Gaussian distribution standard deviation gives the threshold.
Here in our research, the fixing of the significance of the thresholds in either of detectors to screen
99.15% of noise is done. Hence, the likelihood of wrong alarms because of noise will be less than 0.85%.
Normally, the Euclidean detector is considered extra sensitive for variations than compared to the 𝜒 2 -
detector. In case the noise factors are not recognized before, the 𝜒 2 -detector is better because it manages the
soft faults better. Nevertheless, a drawback of the 𝜒 2 -detector compared to the reconfigurable Euclidean
detector is its incompetence to identify a false data injection assault.

Figure 9. DDoS attack for a short period transfer response using reconfigurable Euclidean distance

5.6. Proposed IEEE 9-bus system using RED to detect false data injection attack
Figure 10 depicts a 9-bus structure of IEEE with sensors to observe the state factors and the
estimator for bus 3. The 9-bus structure is replicated using the MATPOWER platform in MATLAB. The
voltages and phases, got by unravelling the 9-bus power structure in MATPOWER, are utilized like factors
of state in the RKF estimator. A related framework could be presumed for every bus in the structure. In order
to understand, merely bus 3 is deliberated. The assault order ℎ𝑎 is produced by the opponent. The sensors
which are there in the bus inform their interpretations to the matching RKF estimators and reconfigurable
euclidean detectors. The positive identification of the False Data Injection attack on bus 3 is depicted in
Figure 11.

Figure 10. Proposed false data injection attack using IEEE 9-bus structure

Int J Pow Elec & Dri Syst, Vol. 13, No. 4, December 2022: 2086-2097
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2095

Figure 11. IEEE 9 bus system used to detect the false data attack

6. CONCLUSION
The proposed method is implemented using reconfigurable Kalman filter, 𝜒 2 detector and
reconfigurable euclidean detector for smart grid system. The proposed system has improved the detection
efficiency of the different types of faults and attacks such as DDoS, FDIA and Random attacks compared to
the conventional methods (0.51%,0.3% and 0.42%). The proposed model improves the security and
controlling capability of smart grid by reducing Euclidean detector noise. With respect simulation analysis, it
shows the proposed method improves detection rate and security compared with conventional methods.
Future scope: The proposed methods is enhanced to detect the faults in smart electric meters in residential
area along with detection of faults and attacks in smart grids.

ACKNOWLEDGEMENTS
The authors would like to thank, SJB Institute of Technology, Bengaluru, JSS Academy of
Technical Education, Bengaluru, Sri Jayachamarajendra College of Engineering, Mysore, Visvesvaraya
Technological University (VTU), Belagavi and Vision Group on Science and Technology (VGST) Karnataka
Fund for Infrastructure strengthening in Science & Technology Level-2 sponsored “Establishment of
Renewable Smart Grid Laboratory” for all the support and encouragement provided by them to take up this
research work and publish this paper.

REFERENCES
[1] S. R. Salkuti, “Artificial fish swarm optimization algorithm for power system state estimation,” Indonesian Journal of Electrical
Engineering and Computer Science, 2020, pp. 1130-1137, doi: 10.11591/ijeecs.v18.i3.pp1130-1137.
[2] Y. B. S. Bri, “Torque estimator using MPPT method for wind turbines,” International Journal of Electrical and Computer
Engineering, 2020, pp. 1208-1219, doi: 10.11591/ijece.v10i2.pp1208-1219.
[3] M. Khalaf, A. Youssef and E. El-Saadany, “Detection of false data injection in automatic generation control systems using Kalman
filter,” IEEE Electrical Power and Energy Conference (EPEC), 2017, pp. 1-6, doi: 10.1109/EPEC.2017.8286194.

Efficient detection of faults and false data injection attacks in smart grid … (Prakyath Dayananda)
2096  ISSN: 2088-8694

[4] M. M. Rana and L. Li, “Distributed Generation Monitoring of Smart Grid Using Accuracy Dependent Kalman Filter with
Communication Systems,” 12th International Conference on Information Technology - New Generations, 2015, pp. 496-500, doi:
10.1109/ITNG.2015.154.
[5] Y. Wang, Z. Zhang, J. Ma and Q. Jin, “KFRNN: An effective false data injection attack detection in smart grid based on Kalman
filter and recurrent neural network,” IEEE Internet of Things Journal, vol. 9, no. 9, pp. 6893-6904, 2022, doi:
10.1109/JIOT.2021.3113900.
[6] Y. Liu and L. Cheng, “Relentless false data injection attacks against kalman filter based detection in smart grid,” IEEE
Transactions on Control of Network Systems, doi: 10.1109/TCNS.2022.3141026.
[7] J. Sawodny, O. Riedel and T. Namerikawa, “Detection of attacks in smart grids via extended Kalman filter and correlation
analysis,” 59th Annual Conference of the Society of Instrument and Control Engineers of Japan (SICE), 2020, pp. 663-669, doi:
10.23919/SICE48898.2020.9240229.
[8] S. H. Mohamad, M. A. M. Radzi, N. F. Mailah, N. I. A. Wahab, A. Jidin and M. Y. Lada, “Adaptive notch filter under indirect and
direct current controls for active power filter,” Bulletin of Electrical Engineering and Informatics, vol. 9, no. 5, pp. 1794-1802,
2020, doi. 10.11591/eei.v9i5.2165.
[9] Y. Zhang, H. Wang and H. Wang, “Integrated navigation positioning algorithm based on improved Kalman Filter,” International
Conference on Smart Grid and Electrical Automation (ICSGEA), 2017, pp. 255-259, doi: 10.1109/ICSGEA.2017.55.
[10] Y. Jiang and Q. Hui, “Kalman filter with diffusion strategies for detecting power grid false data injection attacks,” IEEE
International Conference on Electro Information Technology (EIT), 2017, pp. 254-259, doi: 10.1109/EIT.2017.8053365.
[11] F. Akbarian, A. Ramezani, M. -T. Hamidi-Beheshti and V. Haghighat, “Intrusion detection on critical smart grid infrastructure,”
Smart Grid Conference (SGC), 2018, pp. 1-6, doi: 10.1109/SGC.2018.8777815.
[12] R. Shivaji, k. R. Nataraj, S. Mallikarjunaswamy ank K. R. Rekha, “Implementation of an effective hybrid partial transmit sequence
model for peak to average power ratio in MIMO OFDM system,” ICDSMLA 2020. Lecture Notes in Electrical Engineering,
Springer, vol. 783. PP. 1343-1353, 2020, doi: 10.1007/978-981-16-3690-5_129.
[13] B. Aissa, T. Hamza, G. Yacine, and N. Mohamed, “Impact of sensorless neural direct torque controlin a fuel cell traction system,”
International Journal of Electrical and Computer Engineering, vol. 11, no. 4, pp. 2725-2732, 2021, doi:
10.11591/ijece.v11i4.pp2725-2732.
[14] N. Zhou, D. Meng, Z. Huang and G. Welch, “Dynamic state estimation of a synchronous machine Using PMU data: a comparative
study,” IEEE Transactions on Smart Grid, vol. 6, no. 1, pp. 450-460, 2015, doi: 10.1109/TSG.2014.2345698.
[15] T. N. Manjunath, S. Mallikarjunaswamy, M. Komala, N. Sharmila and K. S. Manu, “An efficient hybrid reconfigurable wind gas
turbine power management system using MPPT algorithm,” International Journal of Power Electronics and Drive Systems, vol.
12, no. 4, pp. 2501-2510, 2021, doi: 10.11591/ijpeds.v12.i4.pp2501-2510.
[16] J. Zhao and L. Mili, “A decentralized h-infinity unscented Kalman filter for dynamic state estimation against uncertainties,” IEEE
Transactions on Smart Grid, vol. 10, no. 5, pp. 4870-4880, 2019, doi: 10.1109/TSG.2018.2870327.
[17] M. M. Rana, “Extended Kalman filter based distributed state estimation algorithm for cyber physical systems,” 3rd International
Conference on Inventive Computation Technologies (ICICT), 2018, pp. 653-655, doi: 10.1109/ICICT43934.2018.9034255.
[18] K. Manandhar, C. Xiaojun, F. Hu and Y. Liu, “Combating false data injection attacks in smart grid using Kalman filter,”
International Conference on Computing, Networking and Communications (ICNC), 2014, pp. 16-20, doi:
10.1109/ICCNC.2014.6785297.
[19] S. R. Salkuti, S. Pagidipala, and S. C. Kim, “Comprehensive analysis of current research trends in energy storage technologies,”
Indonesian Journal of Electrical Engineering and Computer Science, vol. 24, no. 3, pp. 1288-1296,. 2021, doi:
10.11591/ijeecs.v24.i3.pp1288-1296.
[20] S. Mallikarjunaswamy, N. Sharmila, G. K. Siddesh, K. R. Nataraj, M. Komala, “A novel architecture for cluster based false data
injection attack detection and location identification in smart grid. advances in thermofluids and renewable energy,” Lecture Notes
in Mechanical Engineering. Springer, Singapore, 2021, doi: 10.1007/978-981-16-3497-0_48.
[21] S. Mallikarjunaswamy, N. Sharmila, D. M. Kumar, M. Komala and H. N. Mahendra, “Implementation of an effective hybrid
model for islanded microgrid energy management,” Indian Journal of Science and Technology, vol. 13, no. 27, pp. 2733-2746,
2020, doi: 10.17485/IJST/v13i27.982.
[22] T. A. Madhu, M. Komala, V. Rekha, S. Mallikarjunaswamy, N. Sharmila and S. Pooja, “Design of fuzzy logic-controlled hybrid
model for the control of voltage and frequency in microgrid,” Indian Journal of Science and Technology, vol. 13, no. 35, pp. 3612-
3629, 2020, doi: 10.17485/IJST/v13i35.1510.
[23] V. P. Bhuvana, M. Huemer and A. Tonello, “Battery internal state estimation using a mixed Kalman cubature filter,” IEEE
International Conference on Smart Grid Communications (SmartGridComm), 2015, pp. 521-526, doi:
10.1109/SmartGridComm.2015.7436353.
[24] H. Abbes, H. Abid, K. Loukil, M. Abid and A. Toumi, “Fuzzy-based MPPT algorithm implementation on FPGA chip for multi-
channel photovoltaic system,” International Journal of Reconfigurable and Embedded Systems, vol. 11, no. 1, pp. 49-58, 2022,
doi: 10.11591/ijres.v11.i1.pp49-58.
[25] B. Özsoy and M. Göl, “A hybrid state estimation strategy with optimal use of pseudo-measurements,” IEEE PES Innovative Smart
Grid Technologies Conference Europe (ISGT-Europe), 2018, pp. 1-6, doi: 10.1109/ISGTEurope.2018.8571513.
[26] M. M. Rana, M. K. R. Khan and A. Abdelhadi, “IoT architecture for cyber-physical system state estimation using unscented
Kalman Filter,” Second International Conference on Inventive Research in Computing Applications (ICIRCA), 2020, pp. 910-913,
doi: 10.1109/ICIRCA48905.2020.9183350.
[27] I. Onyegbadue, C. Ogbuka and T. Madeume, “Robust least square approach for optimal development of quadratic fuel quantity
function for steam power stations,” Indonesian Journal of Electrical Engineering and Computer Science, vol. 25, no. 2, pp. 732-
740, 2022, doi: 10.11591/ijeecs.v25.i2.pp732-740.

Int J Pow Elec & Dri Syst, Vol. 13, No. 4, December 2022: 2086-2097
Int J Pow Elec & Dri Syst ISSN: 2088-8694  2097

BIOGRAPHIES OF AUTHORS

Prakyath Dayananda has completed BE in EEE at CIT, Tumkur and M.Tech in

CAID at SSIT, Tumkur and secured Gold medal in M.Tech. I had 9 years Teaching experience
in teaching and am currently working as Assistant professor in SJBIT, Bangalore. He can be
contacted at email: [email protected].

Mallikarjunaswamy Srikantaswamy is currently working as an Associate

Professor in Department of Electronics and Communication Engineering at JSS Academy of
Technical Education, Bangalore. He obtained his B. E degree in Telecommunication
Engineering from Visvesvaraya Technological University Belgaum in 2008, M. Tech degree
from Visvesvaraya Technological University Belgaum in 2010 and was awarded Ph. D from
Jain University in 2015.He has 11+ years of teaching experience. His research work has been
published in more than 42 International Journals and conference. He received funds from
different funding agencies. Currently guiding five research scholars in Visvesvaraya
Technological University Belgaum. He can be contacted at email:
[email protected].

Sharmila Nagaraju has completed her B.E in EEE at SJCE, Mysore and M. Tech
in CAID at NIE Mysore. Secured second rank in Bachelor of Engineering degree. She has
Eight years of experience in teaching and is currently working as an Assistant Professor in
RNSIT, Bangalore.electronics and its applications. She can be contacted at email:
[email protected].

Rekha Velluri completed her B.E and M. Tech in Computer Science and
Engineering from Visvesavaraya Technological University Belgavi. She has more than 16
years of teaching experience. Published many papers in national and international conference
and currently working as an Assistant Professor in Christ University. She can be contacted at
email: [email protected].

Doddananjedevaru Mahesh Kumar is presently working as Associate Professor

in the Dept. of Electronics and Instrumentation Engineering, JSS Academy of Technical
Education, Bengaluru. He is working in the teaching field from the past 21 years and has
published more than 30 papers in International Journals, National and International
Conferences. He has 4 patent publications & presently guiding three research scholars
under Visvesvaraya Technological University. His research areas include Biomedical Signal
Processing, Sensors and amp; and Transducers. He can be contacted at email:
[email protected].

Efficient detection of faults and false data injection attacks in smart grid … (Prakyath Dayananda)