Content uploaded by Nasir Ayub
Author content
All content in this area was uploaded by Nasir Ayub on Mar 23, 2022
Content may be subject to copyright.
Earth Worm Optimization for Home Energy
Management System in Smart Grid
Mudabbir ALi1, Samia Abid1, Asad Ghafar1, Nasir Ayub1,Hafsa Arshad1, Sajawal
Khan1and Nadeem Javaid1,∗
Abstract Smart grid based energy management system promises an efficient con-
sumption of electricity. For optimized energy consumption, a bio inspired meta-
heuristic algorithms: Earth Worm Algorithm (EWA) and Bacterial Foraging Algo-
rithm (BFA) are presented in this paper. In this work, we targeted residential area.
Our aim is to reduce the electricity cost and Peak to Average Ratio (PAR). We
have used the Critical Peak Pricing (CPP) scheme for calculating electricity bill.
Through simulations, we have compared the results of EWA, BFA and unscheduled
appliances. After implementing our techniques, EWA based energy management
controller gives more efficient results than BFA in term of cost, while for PAR re-
duction, BFA performs better than EWA.
Keywords
Smart Grid, Meta heuristic techniques, EWA algorithm, BFA algorithm, Critical
Peak Point, Home Energy Management System, PAR
1 Introduction
Energy management is one of the interesting area for scientific research. The de-
mand of energy is increasing with the passage of time due to increased number
of homes, industries and other commercial buildings. Increasing demand of any-
thing requires a manageable production in order to facilitate the consumer with the
best satisfaction. Similarly, the generation of energy also needs to be managed. For
smart management of electricity distribution among consumers, Smart Grids (SGs)
are introduced. The main functions of SG are energy controlling that it gather, dis-
tribution and act on information about the behavior of all participants including
industries, homes and other buildings. Demand Side Management (DSM) facilitates
1COMSATS Institute of Information Technology, Islamabad 44000, Pakistan
∗Correspondence: www.njavaid.com, nadeemjavaidqau@gmail.com
1
2 Mudabbir Ali et al.
more proficient and reliable grid tasks in SG. Moreover, SG increases the connectiv-
ity, coordination and automations between transmission and consumption of energy.
Fig.1 represents the abstract architecture of SG framework. Energy Management
Controller (EMC) is a device, through which we can adjust our appliances in home.
EMC is connected via Wide Area Network (WAN) that communicates with the SG
domain for electricity management. In Home Energy Management System (HEMS)
all the components perform their task, i.e. Smart Meters (SM), EMC, appliances
etc. Different pricing schemes are provided by the utility that are used for calculat-
ing the electricity bills in different scenarios like hourly base, daily base etc. The
main focus of this paper is to reduce electricity cost and PAR for HEMS through the
implementation of optimization techniques.
Rest of the paper is organized as follows: Section 2 contains the brief descrip-
tion of related work. Next, Section 3 explains our proposed solution. Simulation
results are discussed in Section 4. Finally, the conclusion of our work is described
in Section 5 by pointing out the future work.
Fig. 1: Energy Flow with Smart Grid
2 Related work
From decades, number of research have been done globally on SG and energy man-
agement system. Different optimization techniques have a key role in scheduling of
appliances in order to manage electricity consumption. Different researchers pro-
posed different types of technique on energy management, i.e. heuristic algorithms
Earth Worm Optimization for Home Energy Management System in Smart Grid 3
based, dynamic programming algorithms, divide and conquer technique based etc.
In this regard, few research papers are discussed below.
In [1], EMC general architecture for Home Area Network (HAN) is proposed.
Demand Response (DR) is used in SG to indicate the electricity price in real time
and to transform it to EMC. Through Real Time Pricing (RTP) scheme it is not
possible to handle PAR balancing, therefore authors integrated RTP and Inclined
Block Rate (IBR) pricing scheme to reduce PAR. Genetic Algorithm (GA) is used
as optimization technique for scheduling the appliances. The main target is to re-
duce electricity cost and PAR. Authors successfully reduced electricity cost after
implementing their technique. They also reduce the delay time of appliances’ op-
eration which means User Comfort (UC) has also taken under account. So, they
proved that integrated technique (IBR and RTP) provides better results than alone
RTP. According to paper, security issues are not addressed. Authors used five heuris-
tic techniques in [2], and compared the results in order to achieve the optimal
solution for HEMS. The mentioned techniques are GA, Binary Particle Swarm
Optimization (BPSO), Bacterial Foraging Optimization Algorithm (BFOA),Wind-
Driven Optimization (WDO) algorithm and their proposed Hybrid Genetic Wind-
Driven (GWD) algorithm. Through statistical analysis, results depicted that Hybrid
GWD reduces cost 10% percent by GA and 33% by WDO while rest of other results
are far behind from their proposed technique. So, overall Hybrid GWD results are
better in reducing bill as compared to GA, BPSO, BFOA and WDO. The limitations
are that the priority based scheduling is not intelligently installed and did not com-
pare with other techniques for better exploration.
In [3], authors have presented an EMC for avoiding the peak formation, the elec-
tricity bill reduction and maintaining the acceptable UC by formulating problem as
Multiple Knapsack Problem (MKP). Authors also integrated Renewable Energy Re-
sources (RES) in HEMS. They used two pricing schemes for bill calculation; Time
Of Use (TOU) and IBR. In paper three heuristic algorithms GA, BPSO, and Ant
Colony Optimization (ACO) are used to achieve the optimal solution for their de-
signed model. The results show the validity of the proposed solution that it reduces
the electricity bill and PAR. As there is a trade-off between cost reduction and UC,
UC is compromised. New meta-heuristic technique inspired by the nature of earth-
worm reproduction system is proposed in [4]. That reproduction is categorized into
2 types (Reproduction 1 and Reproduction 2). Reproduction 2 gives the optimized
solution after applying 9 crossovers functions then they applied cauchy mutation
for the extraction of most optimized value. Authors just designed the technique in
the paper but did not apply on SG for its performance exploration. According to
[5], DSM is introduced to handle the issue of energy demands plus to utilize the
energy efficiently. Authors proposed the hybrid technique and compared their pro-
posed technique with GA, BFA and unscheduled appliances. After simulation and
practical implementation, they successfully achieved the goal they want by reducing
electricity cost and maintained UC with the comparison of unscheduled appliances
but, still limited to some extent as BFA shows better result in term of cost reduction.
In [6], authors focused on scheduling of load and power trading with RES. They
have adopted dynamic programming scheme to schedule operation of appliances
4 Mudabbir Ali et al.
and game theoretic approach to interact between different users with excess energy
generation. Users are encouraged to avoid reverse transferring of excess energy to
utility using RES. After simulations load is scheduled, electricity cost is reduced
and user with excess energy generation is able to sell the electricity to local users
cheaper than energy provider. As a result there is a reduction in electricity payments
and proposed system provides the gateway to earning revenues. Moreover, no secu-
rity issue is mentioned regarding to protection of RES for trading the excess energy
transmission.
The consumption of energy is managed in residential SG network according to [7].
Each house is configured by 2 types of demands, flexible and essential demands, the
subcategory of flexible demands are delay-sensitive and delay-tolerant demands.
Service of delay-sensitive demand is considered to be more important than delay-
tolerant demands. Meanwhile, to decrease the waiting time of delay-tolerant de-
mands, these demands need to be upgraded for high priority queue. A centralized
algorithm based on dynamic programming is proposed by authors to optimize the
solution. As an addition, a distributed algorithm is proposed for implementation and
the neural network. Simulation results show that the usage of electricity in house-
hold is managed and for flexible demands operation delays are reduced as well.
There is an issue regarding to inconsideration of parameters tuning.
Authors in [8], target the Peak Power Consumption (PPC) minimization. To ad-
dress such issue, authors proposed the family plan approach that divides users into
schedule groups. User’s appliances are scheduled for PPC minimization through
clustering scheme. Their aim is to achieve the reduction in payments, PPC and fuel
cost. PPC is reduced by scheduling the jobs of each controllable family. They also
balance the PAR and PPC by incorporating divide and conquer rule. User has to
stay connected as a group member as mentioned in paper. Although their simula-
tion work well however, limitations are still for research analysis that their system
cannot handle interruptible jobs. In [9], user aware management approach is used
that manages load by user preferences. Problem is trade-off between cost and UC.
So, Game theoretic based energy optimization technique is used. It is different from
other demand management systems, that it allows the user to prioritize the user pref-
erences and savings. Load is distributed by means of game theoretic approach which
has successfully reduced the electricity cost. Authors also analyze the performance
of algorithm that approaches to 96% closer to optimal solution. As limitations, their
proposed technique is not perfect in the presence of multiple energy resources.
Integer Linear Programming (ILP) technique is proposed in [10], for minimizing
the peak load and scheduling the load in SG. Moreover, another aim is to sched-
ule the power for power shift-able appliances and to optimize the operation time
for time shift-able appliances. Authors are able to schedule both optimal operation
time, optimal power and accomplished effective electricity consumption. Again, au-
thors did not address the UC issue. Authors proposed a power scheduling strategy
for the residential consumers in [11], to achieve a trade-off between the electricity
cost and discomfort. Day ahead pricing signal is used in this paper. The result shows
that the authors achieved low electricity price, reduction in PAR and achieved the
limited user comfort because of the trade-off between cost and user comfort. As lim-
Earth Worm Optimization for Home Energy Management System in Smart Grid 5
itation for this paper authors did not provide incentive based DR load adjustments.
As [12], elaborates that DR is essential for reducing electricity costs and loads. Au-
thors focused on developing strategy for Heating, Ventilating and Air-Conditioning
(HVAC) to respond to real time pricing for reduction of peak load. Proposed tech-
nique Dynamic Demand Response (DDR) changes the temperature’s set point to
control HVAC loads depending on electricity retail price after each 15 minute then
shift the loads of appliances partially. When DDR changed the set point of tem-
perature, discomfort level decreases. Simulation showed DDR in residential HVAC
systems can reduce peak load and electricity bill. Electricity cost for both hot and
cold months are reduced to 31% and 29% respectively.
3 Proposed Solution
In residential areas, every home is considerably being equipped with SM that pro-
vide two-way communication between customer and utility, shown in Fig. 2. In our
proposed system, we have categorized 13 appliances into 3 classifications; inter-
ruptible, non-interruptible and fixed appliances. We have scheduled the appliances
according to [6]. Classified appliances are mentioned below in Table1 .
Table 1: Classification of Appliances
Non-Interruptable Interruptable Fixed
Washing Machine Air Conditioner Light
Dish Washer Refrigerator Fans
Cloth Dryer Water heater Cloth iron
Space heater Microwave oven
Toaster
Coffee maker
Interruptible appliances are those, that can be turned off any time during opera-
tion. Non-interruptible appliances, that cannot be interrupted during operation time,
while Fixed appliances are considered as non-manageable or non-modified appli-
ances because they are regular appliances that cannot affect much on load schedul-
ing issues. End users are allowed to customize the parameters of appliances ac-
cording to their will by interacting with EMC to establish connection with utility.
Moreover, end users are remained updated about the pricing scheme announced by
the utility through DSM. The aim of utility is to educate users about the efficient
energy consumption. Fig.2 elaborates the generic work flow of DSM for HEMS.
Peak and off peak hours are the time slots, where load of energy varies during the
whole day. In this work we have assumed that 11 to 18 hours are the slots for peak
hours during the whole day. Hence, the parameters are initialized in advance by end
user using DSM. Parameters for non-interruptible appliances are shown in Table2,
6 Mudabbir Ali et al.
Fig. 2: DSM Work Flow
Table3 for interruptable appliances and Table4 represents the parameters for fixed
appliances.
Table 2: Non-Interruptable Appliances
Appliances α β γ Ω (kWH)
Washing Machine 8 16 5 0.78
Dish washer 7 12 5 3.60
Cloth dryer 6 18 5 4.40
Table 3: Interruptable Appliances
Appliances α β Ω (kWH)
Air conditioner 6 24 1.44
Refrigerator 6 24 0.73
Water heater 6 24 4.45
Space heater 6 24 1.50
According to Table2, αrepresents the starting time for the appliances, βrepre-
sents ending time of appliances in hours. For specific appliance, γrepresents the
number of operational hours in which energy is utilized at any time slot between
αand β, while Ωrepresents the load of specific appliance in kWH. Similarly, in
Table3 for interruptible appliances, α,βand Ωare same as described for Table2
however, there is no constraint of operational hours in which appliances have to op-
Earth Worm Optimization for Home Energy Management System in Smart Grid 7
Table 4: Fixed Appliances
Appliances Ω(kWH)
Lighting 0.6
Fans 0.75
Clothes iron 1.5
Microwave oven 1.18
Toaster 0.5
Coffee maker 0.8
erate for limited hours between αand β. On the other hand in Table4, parameters for
fixed appliances are initialized in which appliances can be operated independently
at any hour during a whole day.
We have used CPP scheme for electricity bill calculation. The main purpose to use
this pricing scheme is that it provides the accurate information to customer about the
cost of electricity, so that customer can make decisions that how and when to use
electricity. CPP pricing scheme awares the customer about the peak hours during the
day. CPP rate offers lower prices during all other times. Therefore, the customer can
have the opportunity of better assessement and thus reduces the overall energy costs.
A bio-inspired meta-heuristic algorithm; EWA [4], is used for optimizing the elec-
tricity consumption. In EWA, there are two kinds of reproductions; Reproduction
1 and Reproduction 2. In nature, Reproduction 1 generates only 1 offspring either
from male or female, while Reproduction 2 may generates more than one offspring
at a time. Multiple crossover operators are used in order to improve the version of
crossover head, addition to this Cauchy mutation is implemented to extract the best
value after iterations. An abstract representation of natural behavior of earthworm
is mentioned in Fig. 3.
Fig. 3: Earth Worm Natural Behaviour.
8 Mudabbir Ali et al.
3.1 EWA Algorithm
The reproduction quality of earthworm performs multiple optimization steps, pro-
duction steps of earthworm is optimized by the following scenarios:
•Every earthworm in population has the ability to regenerate its offspring in na-
ture and ever individual earthworm has the capability of 2 reproductions.
•Every child of earthworm singular generated holds the entire genetic factor
whose length is equivalent to parental earthworm.
•The earthworm singular with the fitness permits on straight next generation,
and can not be altered by operators. This can be an assurance that population of
earthworm can not fail in the increment in generations.
Steps of EWA algorithm are mentioned in Algorithm 1
Algorithm 1 EWA
1: procedure START
2: Initialization: Generates counter of t = 1; Set P as population of NP individual earthworm
which is randomly distributed in search space; numbers of kept earthworm are set as nKEW,
maximum generation MaxGn,αas similarity factor, proportional aspect β, constant γ= 0.9.
3: Evaluation of Fitness: each earthworm is evaluated individually according to its position
4: While till best solution is not achieved or t <MaxGen
5: All the earthworms in population are then sorted according to their fitness values
6: for i = 1 to NP (all earthworms) do
7: Generate offspring xi1 through Reproduction 1
8: Generate offspring through Reproduction 2
9: Do crossover
10: if i <nKEW then
11: set the number of particular parents (N) and the produced off springs (M); Select the N parents
using method i.e. roulette wheel selection; Generate the M offspring; Calculating xi2 according
to offspring M generated
12: else
13: Randomly an individual earthworm as xi2
14: end if
15: Update the location of earthworm end for i
16: for j = nKEW + 1 to NP (earthworm individuals non-kept)
17: do Cauchy mutation
18: end for j
19: Calculate the population according to the newly restructured positions;
20: t = t + 1.
21: Step 4:
22: end while
23: Step 5:
24: Best solution is extracted
25: End.
Our next task is to compare EWA with BFA [5]. For BFA, we have used same
appliances classification and parameters as mentioned in Table 3, Table 4 and Table
5. Then we merged both techniques in order to depict the clear justifications. BFA
Earth Worm Optimization for Home Energy Management System in Smart Grid 9
is also a bio-inspired technique in which animals with poor foraging strategies are
eliminated and replaced by healthy ones in nature. The main focus of this algorithm
is to allow cells to randomly transmit towards the optimal solution by means of
chemotaxis, swarming, reproduction and elimination-dispersal steps. Such behavior
involve different steps as shown in Algorithm 2.
Algorithm 2 BFA
1: procedure START
2: Step 1:
3: Chemotaxis: number of chemotactic steps measures the length of bacteria’s life time.
4: if favorable environment = true
5: bacteria = continue swimming
6: else
7: bacteria = direction changed
8: do swarming
9: bacterias attraction complete.
10: Step 2:
11: Reproduction: for optimized value
12: Calculate the reproduction of bacteria quality
13: then bacteria reproduce the generation
14: fitness value = found
15: creat next generation.
16: Step 3:
17: Elimination-dispersal: Cells are eliminated and new random samples are inserted for local
search
18: reproduction created for chemotaxis
19: End.
4 Simulation Results
For simulations, we have evaluated the performance of different techniques in MAT-
LAB. Scheduling of appliances are managed according to [6]. Different parameters
of appliances can be defined through customers also. We have conducted the simu-
lations on the basis of energy consumption, peak to average ratio, waiting time and
electricity cost.
In Fig 4, the numbers of hour are mentioned on the x-axis and load (kWhs) of
electricity is given on y axis of graph. After simulations, the load is maintained for
BFA scheduled appliances between 1 to 6 hours. BFA shows almost the constant
load of 2kWh at 1 to 6. While EWA permits a little bit higher load between same
time slots (1 to 6) that is almost 3 kWhs. On the other hand unscheduled appliances
are showing the consumption of electricity slightly higher than BFA at 1 to 3 hours
but less than EWA during the time slots of 3 to 5 hours. Load of BFA gradually
increases during time slots of 6 to 8 with the variant resultant load of 6 to 17 kWHs.
For EWA, in 6 to 10 hours, load increases from 3 to 18 kWHs. On the other hand
unscheduled appliances also depict the same behavior with different results having
10 Mudabbir Ali et al.
maximum load of 19 kWHs at 8 hour. After 11 hour, load of appliances starts de-
creasing for EWA and BFA while for unscheduled appliances, load decreases after
8 hour. We can not totally neglect the consumption of electricity in peak hours how-
ever; we at least accomplished to optimize the consumption to some extent using
BFA and EWA algorithms. In case of unscheduled appliances, we can clearly visu-
alize that the consumption is not smart. In peak hours, load is higher at peak hours
because no smart technique is used. We can see the effect of shifting the loads that
in peak hours, unscheduled appliances are not efficient while our proposed scheme
works well, another reason is that we have classified different appliances into three
categories. Overall, the loads of appliances are same during 24 hours but the con-
sumption of energy is optimized in order to avoid higher electricity cost.
Time (hours)
1 2 3 4 5 6 7 8 9101112131415161718192021222324
Load (kWh)
0
5
10
15
20
Unscheduled
EWA Scheduled
BFA Scheduled
Fig. 4: Electricity Load for 24 Hours.
As PAR is concerned, we have utilized our techniques for optimization. PAR is
reduced which is beneficial to balance the power supply and demand. In case of PAR
reduction we have given more priority to interruptible and non- interruptible appli-
ances than fixed appliances. In Fig 5, we can see that there is a difference regarding
the PAR reduction, that EWA technique reduces PAR to almost 5% and BFA to 7%
as compared to non scheduled appliances. So therefore, regarding to PAR reduc-
tion BFA performs better than EWA. For avoiding peak formation, we have used
CPP pricing scheme that provides information to user about rates of electricity bills.
Hence reduction in electricity bills are accomplished. Moreover, EWA algorithm’s
quality of multiple crossovers assists to achieve best solution for optimizing the
problem. Unscheduled appliances show poor results because, they do not tackle the
peak formation problems while our scheduled schemes are designed to avoid peak
Earth Worm Optimization for Home Energy Management System in Smart Grid 11
formation in any hour of the day because appliances are optimally distributed for 24
hours.
Unscheduled EWA Scheduled BFA Scheduled
PAR
0
2
4
6
8
10
12
14 Peak to Average Ratio
Fig. 5: Peak to Average Ratio.
Time (hours)
1 2 3 4 5 6 7 8 9101112131415161718192021222324
Electricity Cost (cent)
0
500
1000
1500
2000
Unscheduled
EWA Scheduled
BFA Scheduled
Fig. 6: Electricity Cost.
Fig 6 elaborates that the result of scheduled appliances are optimal as compared
to unscheduled appliances in term of electricity cost reduction. For 24 hours, maxi-
mum electricity bill in case of unscheduled appliances are 1998 cents at peak hours
12 Mudabbir Ali et al.
Unscheduled EWA Scheduled BFA Scheduled
Total Cost
0
1000
2000
3000
4000
5000
6000
7000
Fig. 7: Total Electricity Bill.
between 11 to 12 hours, while for BFA scheduled appliances we can see that during
11 to 12 hours the maximum cost lies almost at 1200 cents and in case of EWA
technique, the results are more efficient. During peak hours, the cost is reduced to
450 cents in same time slots (11 to 12). From simulation results, we can extract
the information in chunks that, during 1 to 6 hours both scheduled and unscheduled
appliances have almost same electricity cost. When we slightly go further from 7
to 17 hours we can see the drastic change that our proposed optimization technique
EWA is forming very little peak in peak hours, i.e. 200 to 450 then 450 to 130 cents.
On the other hand BFA shows slightly higher peak formation as compared to EWA
but less then unscheduled. Such accomplishment was achieved due to the BFA pa-
rameters (reproduction and elimination) and EWA’s two reproduction formations.
Then total electricity bill reduction is mentioned in Fig 7 to almost 45% for BFA
and 55% for EWA, the fact is that we have delayed the operation of some appliances
according to designed objective functions.
Waiting time of appliances in Fig 8 directly affects the cost of electricity bill and
User Comfort. Waiting time is increased because we have reduced the electricity
cost by shifting the load of appliances from peak to off peak hours. According to
Fig 8 BFA is better than EWA in term of waiting time. Waiting time factor has an
important role for user satisfaction, increasing waiting time User Comfort also sac-
rifices. The relationship between Electricity cost and waiting time is inversely por-
tion. So, in Fig 8 simulation, waiting time for EWA has increased but bill decreased.
EWA has sacrificed the user comfort and user has to wait for specific appliances to
be operated.
Earth Worm Optimization for Home Energy Management System in Smart Grid 13
BFA Scheduled EWA Scheduled
waiting time
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5waiting time
Fig. 8: Waiting Time.
5 Conclusion
We have proposed the techniques for optimized electricity consumption in HEMS.
Through simulations we have deduced the results that EWA performs better than
BFA in term of cost and in term of PAR reduction BFA have a better results. Using
CPP pricing scheme PAR reduction is achieved. We are able to accomplish the op-
timal results after shifting the load of appliances from peak hours to off peak hours.
As there is a trade-off between cost reduction and UC, therefore in our work we have
compromised the UC. In future we will focus on reducing the waiting time to meet
the acceptable UC and concentrate on contribution of Renewable energy resources.
References
1. Zhao, Z., Lee, W. C., Shin, Y., Song, K. B. (2013).An optimal power scheduling method for
demand response in home energy management system. IEEE Transactions on Smart Grid, 4(3),
1391-1400.
2. Javaid, N., Javaid, S., Abdul, W., Ahmed, I., Almogren, A., Alamri, A. and Niaz, I. A. (2017).A
hybrid genetic wind driven heuristic optimization algorithm for demand side management in
smart grid.Energies, 10(3), 319.
3. Rahim, S., Javaid, N., Ahmad, A., Khan, S. A., Khan, Z. A., Alrajeh, N., Qasim, U. (2016).Ex-
ploiting heuristic algorithms to efficiently utilize energy management controllers with renew-
able energy sources. Energy and Buildings, 129, 452-470.
4. G.G. Wang, S. Deb, L.D.S. Coelho, Earthworm optimization algorithm: a bio-inspired meta-
heuristic algorithm for global optimization problems, Int. J. Bio-Inspired Comput. (2015).
14 Mudabbir Ali et al.
5. Adia, Nadeem javaid, Mateen, zahoor Ali and Umer Qasim, Demand Side Management Using
Hybrid Bacterial Foraging and Genetic Algorithm Optimization Techniques, ", IEEE TRANS-
ACTIONS ON SMART GRID, conference paper, 6-8 July 2016, Japan
6. Samadi, Pedram, Vincent WS Wong, and Robert Schober. ”Load scheduling and power trading
in systems with high penetration of renewable energy resources.” IEEE Transactions on Smart
Grid 7.4 (2016): 1802-1812.
7. Yi Liu, Chau Yuen, Rong Yu, Yan Zhang, and Shengli Xie, " Queuing-Based Energy Consump-
tion Management for Heterogeneous Residential Demands in Smart Grid ", IEEE TRANSAC-
TIONS ON SMART GRID
8. Qiuyuan Huang, Xin Li, Jing Zhao, Dapeng Wu, Fellow, IEEE, and Xiang-Yang Li,"Social
etworking Reduces Peak Power Consumption in Smart Grid", IEEE TRANSACTIONS ON
SMART GRID, VOL. 6, NO. 3, MAY 2015
9. Naouar Yaagoubi, and Hussein T. Mouftah," User-Aware Game Theoretic approach for De-
mand Management", IEEE TRANSACTIONS ON SMART GRID
10. Z. Zhu, J. Tang, S. Lambotharan, W. H. Chin, and Z. Fan, “A n integer linear programming
based optimization for home demand-side management in smart grid,” in Innovative Smart Grid
Technologies (ISGT), 2012 IEEE PES, pp. 1–5, IEEE, 2012.
11. K. Ma, T. Yao, J. Yang, and X. Guan, “Residential power scheduling for demand response
in smart grid,” International Journal of Electrical Power and Energy Systems, vol. 78, pp.
320–325, 2016.
12. Ji Hoon Yoon, StudentMember, IEEE, Ross Baldick, Fellow, IEEE, and Atila Novoselac. "Dy-
namic Demand Response Controller Based on Real-Time Retail Price for Residential Buildings
", IEEE TRANSACTIONS ON SMART GRID, VOL. 5, NO. 1, JANUARY 2014
