Residential Energy Consumption Controlling
Techniques to Enable Autonomous Demand Side
Management in Future Smart Grid Communications
M. N. Ullah1,N.Javaid
1Dept of Electrical Engineering, COMSATS Institute of IT, Islamabad, Pakistan.
2CAST, COMSATS Institute of IT, Islamabad, Pakistan.
Abstract—This paper presents an overview of home appliances
scheduling techniques to implement demand side management in
smart grid. Increasing demand of consumers have affected the
power system badly as power generation system faces a number
of challenges both in quality and quantity. Economical generation
and efﬁcient consumption can solve this problem in future smart
grid as it is integrated with information and communication
technologies. Smart grid has opportunities to employ different
pricing schemes which help also in increasing the efﬁciency of
appliances scheduling techniques. Optimal energy consumption
scheduling minimizes the energy consumption cost and reduces
the Peak-to-Average Ratio (PAR) as well as peak load demand.
In this work, we discuss different energy consumption scheduling
schemes that schedule the household appliances in real-time to
achieve minimum energy consumption cost and reduce peak
demand to shape the load curve.
Index Terms—Demand side management, optimal energy con-
sumption scheduling, peak load demand, smart grid.
A system that implements communication and information
technology in electrical grid is known as smart grid. Smart grid
system gathers information about the activities of electrical
energy suppliers and users. Smart grid improves the cus-
tomers’ load utilization by deploying the communication based
monitoring and controlling architectures . With the addition
of different types of new loads e.g. Plug-in Hybrid Electric
Vehicles (PHEVs), the normal residential load has potentially
increased. Hence it is very important to develop new methods
for peak load reduction. There are some environmental issues
related to current power systems. Some countries widely use
the oil and coal ﬁred power plants to meet the peak demands,
as a result a huge amount of CO2and green house gases is
emitted. Smart grid enables Demand Side Management (DSM)
to overcome these problems. DSM was proposed in the late
1970s . DSM monitors, plans and implements those utility
functionalities that are designed to encourage low consumption
by the consumers during peak hours in order to shape the
utility load curve. DSM programs are implemented to exploit
better utilization of current available generating power capacity
without installing new power generation infrastructure .
DSM programs facilitate users to shift loads from peak hours
to off peak hours to reduce the peak load . Worldwide
energy utilization in buildings is approximately 40% of global
power consumption . Currently, consumption of electricity
in buildings is not efﬁcient and is leading towards the wastage
of billions of dollars and huge amount of green house gas
emission. Better utilization of energy consumption in buildings
is an important issue. Heavy loads run in the peak hours
due to which utility load curve potentially goes high. DSM
controls the residential loads by shifting the load from peak
hours to off-peak hours in order to reduce the peak load
curve and improve energy efﬁciency by scheduling the energy
In previous literature, various load management techniques
have been discussed to enable autonomous DSM in future
smart grid. Incentive based energy consumption controlling
schemes are discussed in , , , have explained a Direct
Load Control (DLC) scheme for residential load control to
enable demand side management. In , a home energy
consumption scheduling technique is elaborated which uses
Energy Management Controllers (EMCs) for scheduling the
appliances. , has presented a priority based scheduling
scheme, in which the appliance that has higher priority accord-
ing to load curve, switched ON ﬁrst without any restriction.
Low priority devices are switched ON with some delay.
Different home energy management schemes in future smart
grid are discussed in . The home energy management
schemes are combined with different pricing schemes in order
to make the schemes more efﬁcient e.g. a day ahead pricing
has been used in energy management scheme to minimize the
electricity charges of a consumer .
Our work focus on different energy consumption scheduling
schemes to meet the peak load demand and to reduce the
monetary cost. These residential load controlling schemes
are based on optimization and different scheduling criteria.
Scheduling of appliances involves different pricing schemes.
By scheduling the appliances, heavy loads are shifted from
peak-hours to off-peak hours according to their energy con-
sumption proﬁle. In this way, peak-hours load is reduced by
controlling the appliances.
Rest of the paper is organized as follows: Section II de-
scribes different scheduling schemes for demand side man-
agement. In Section III, we conclude this paper.
II. Different Scheduling Schemes for DSM
Efﬁciency of power consumption is an important factor.
Now a days customers expectations are increasing both in
2013 Eighth International Conference on Broadband, Wireless Computing, Communication and Applications
978-0-7695-5093-0/13 $31.00 © 2013 IEEE
2013 Eighth International Conference on Broadband, Wireless Computing, Communication and Applications
978-0-7695-5093-0/13 $31.00 © 2013 IEEE
quality and quantity. Due to limited energy assets and ex-
pensive process of integrating new energy resources, there
is an important need to improve our system power utiliza-
tion. Energy efﬁciency refers to using minimum energy to
provide the same or improved level of service to the energy
consumer in an economically efﬁcient way. It includes using
minimum energy at any time, including during peak periods.
Smart grid implements advance metering infrastructure for
the load prediction in residential areas and improves energy
efﬁciency . To achieve high reliability in electric grid
utility companies need to reduce the peak load demand. To
avoid the peak condition in the grid, domestic and industrial
devices such as air conditioners, refrigerators and heater are
shifted to off peak hours.
In future smart grid, both power companies and users can
take advantage from economical and environmental aspects of
smart pricing models . Consumer’s energy demand increase
during peak hours. Utility companies run peaker power plants
to meet this peak demand during peak hours. Peaker plants
charge higher prices per kilo watt hour. Communication and
metering technologies enable smart devices in homes to reduce
peak load demand during high cost peak hours. Electricity
prices are high during peak hours and low during off peak
hours. Therefore consumers avoid high price peak hours and
shift their heavy loads to off peak hours. General approach
for the management of energy consumption in buildings is
shifting consumption and reducing consumption. Smart grid
applies DSM to cope with energy consumption cost. Energy
consumption cost can be minimized in homes by shifting the
high load from peak hours to off peak hours.
In previous literature, following techniques have been pro-
posed to minimize the peak load in peak hours and monetary
A. An Autonomous Three Layered Structure Model for
In , a scheduling model has been presented in which a
layered structure consisting of three modules for Admission
Control (AC), Load Balancer (LB) and Demand Response
Manager (DRM) are used to control the peak load demand.
The model of this scheme is shown in Fig.1. Loads are
divided into three different categories based on their load
characteristics for the minimization of peak load demand.
Baseline load is the power consumption of those appliances
that must be run immediately at any time. Baseline load
includes lighting, cooking rang and networking devices etc.
Burst load is the load that is ON for a ﬁxed duration
and starts and stops within the given dead line e.g., washing
machine, dish washer and dryer.
Regular load is related to appliances that are always in
running state during a long period, such as refrigerator, water
heater and Heating Ventilation and Air Conditioning) (HVAC)
The key role of run-time scheduling of appliances is to meet
the power capacity limit while satisfying different capacity cri-
teria e.g. comfort level of domestic users. Present architecture
controls the appliances using online scheduling approach in
the run-time manner.
Fig. 1. Three layered architecture for load management on demand side
In previous literature, many algorithms have been used for
run time scheduling, most popular are Least Slack Time (LST),
Earliest Deadline First (EDF), Bratley and Spring. Spring
algorithm is used for scheduling in AC module. AC manages
the requests at run time coming from smart appliances and
information received from smart meter. AC module is time
triggered and performs adequate load scheduling. It accepts
some requests based on priority (available capacity and power
request) and rejects the rest. The appliances, whose requests
are accepted by AC, start their operation immediately while
rejected requests of appliances are forwarded to LB. The
search space for scheduling the appliances in different time
slots is driven by heuristic function H. At each stage of search
space, the heuristic function H is applied to tasks one by one
that has to be scheduled. The task with least value speciﬁed
by heuristic function called heuristic value is elected to further
extend the current schedule. In this case the heuristic value is
scaled between 0 & 1, which represents the priority of the task.
For example, appliances such as water heater and refrigerator
will have 1 heuristic value when the desired temperature is at-
tained and 0 when the temperature is within the speciﬁed limit
called comfort zone. LB schedules the tasks whose requests
are rejected from AC. LB evenly distributes the electrical load
of appliances over a time frame and schedules the requests
that have been refused by AC. To minimize the total energy
cost, a Mixed Integer Programming (MIP) problem is solved.
Therefore LB minimizes the cost function analogous to energy
price. In different circumstances of considered appliances,
LB establishes an appropriate schedule that would evenly
distribute the appliances load over a time horizon. In such
a way, AC and LB schedule the appliances on run time with
respect to limited capacity constraints and overall peak load
and energy consumption cost is minimized.
B. Backtracking-based Technique for Load Control
A task model is presented in  to schedule the home
appliances for reducing the local peak load as well as the
global peak for effective DSM. The task model is shown in
Fig. 2. Power scheduler operation
Presented scheme designs a power scheduler that is capable
of minimizing peak load in buildings. This model consists of
actuation time, operation length, dead line and consumption
The proposed task model schedules the appliances in real
time. Electric devices are designed as tasks having starting
time, execution time and deadlines. Task Tican be designed by
following parameters: Fishows that whether Tihas preemptive
operation or non-preemptive. Aiis activation time of task, Di
is deadline and Uiillustrates the operation length. Operation of
non-preemptive task can start from its activation time Aito the
latest start time (Di-Ui). For preemptive task, device activation
time must be picked from Uiout of (Di-Ai) time slots.
Backtracking optimization technique is used for scheduling
the appliances. Backtracking additionally frame a search tree
on the allocation table. Scheduler copies the proﬁle entry
of different appliances one by one according to task proﬁle
to the allocation table. This potential search tree consists of
all feasible solutions including worthless solutions. At each
intervening node, which passes to a feasible solution, it checks
whether the node can guide to a feasible solution. If it cannot,
remaining sub tree is reduced. Otherwise iteration proceeds
to the next level. By this phenomenon, scheduler searches
the feasible time slots to schedule the appliances. Optimal
scheduling of appliances reduces the peak load curve and also
reduces the energy consumption cost. This model reduces the
peak load up to 23.1%.
C. Game-Theoretic Based DSM
In , a game-theoretic model based optimization tech-
nique is discussed to schedule the energy consumption of
appliances. In this model users are players and their strate-
gies are their daily scheduling loads. Optimal performance
in terms of energy cost minimization is achieved at Nash
equilibrium of energy scheduling game. The model considers
a common scenario where a single utility company serves
different users. This model systematically manages the appli-
ance schedule and shifts them in order to reduce energy cost.
Energy Consumption Scheduler (ECS) is deployed in smart
meters for scheduling the household appliances. ECS uses a
distributed algorithm to ﬁnd the feasible schedule for each
user. A smart power system with only one energy source and
multiple consumers has been assumed. This scheme considers
two types of appliances, shiftable and non-shiftable. Shiftable
appliances e.g., PHEVs, washing machine and dryer. Non-
shiftable appliances are those that are always in operating
state for a long time period e.g. fridge and lights. ECS only
schedules the shiftable appliances. The operation of ECS is
presented in Fig.3 . In this technique, scheduler manages
Fig. 3. Model with ECS devises deployment
and shifts the appliances energy consumption for appropriate
scheduling. Consider each user n∈N,letAndenote set
of appliances. An energy consumption scheduling vector for
appliance a∈Ancan be deﬁned as:
n,a : Energy consumption of appliance a scheduled for
1hour from user n.
ECS function decides optimal choice for energy
consumption vector Xn,a. Each appliance is scheduled
according to its daily predetermined energy consumption that
n,a =En,a (2)
n,a =0,∀h∈H\Hn,a (3)
En,a: Predetermined daily energy consumption of
αn,a: Interval starting time that appliance consumption can
βn,a: Interval end time that appliance can be scheduled.
Appliance minimum standby power level is deﬁned
n,a and maximum power level by γmax
n,a . Finally feasible
scheduling set for the appliances of user n is acquired as
This feasible schedule is only valid for xn∈χn. Simulation
results show that with the deployment of ECS function in
smart meters, PAR reduces up to 17% and cost reduces up
to 18% as presented in Fig.4 and Fig.5 .
Fig. 4. When ECS not deployed (PAR is 2.1 and the total daily cost is
Fig. 5. When ECS not deployed (PAR is 1.8 and the total daily cost is
D. ECS Device Based Scheduling
In , an energy consumption scheduling (ECS) device
based scheduling technique is discussed. In this technique,
authors consider a scenario of power system where an energy
source (e.g., a generator is connected to electric grid) is shared
by different users. Each user is equipped with an ECS device in
the smart meter shown in Fig.8 . Assume that each user
Fig. 6. Smart grid system model with N load subscribers
is facilitated with smart meter. In this scheme, deployment
of ECS devices in smart meters enables autonomous DSM.
ECS devices are equipped in smart meters connected with
each other. These devices are also connected with power grid
and local area network to communicate with the smart grid
infrastructure. Distributed algorithm is used to schedule the
optimal energy consumption for each subscriber. According
to individual energy needs of all subscribers ECS devices
schedule the energy consumption of household appliances.
ECS devices are interrelating automatically by running an
algorithm to ﬁnd an optimal schedule for energy consumption
of each subscriber. This algorithm reduces the total energy
cost and shape the load curve in peak hours. By subscribing
this algorithm in smart meters, all users pay minimum amount
of utility bills to the utility company as shown in Fig.6 .
A new pricing scheme is also introduced for this model
which is developed from game theoretic model, to reduce
the total cost. Simulation results are demonstrated in Fig.7
and Fig.8 . Simulation results show that ECS devices
efﬁciently schedule the appliances energy consumption in the
Fig. 7. Daily cost $86.47 (ECS devices are not used)
Fig. 8. Daily cost $53.81 (ECS devices are deployed)
E. An Optimal and Autonomous Residential Load Control
Smart pricing models in future smart grid can potentially
beneﬁt both users and utility companies regarding the econom-
ical and environmental advantages. In , an automatic and
optimal energy consumption scheduling scheme is discussed
to minimize the PAR and reduce the waiting time of each
appliance operation in household. For optimal scheduling of
appliances, residential load controller requires the capability
to predict the prices in real time. In this scenario, real-time
pricing and inclining block rates are combined to balance
the load and minimize PAR. An ECS device is deployed
in residential smart meters to control the load of household
appliances. Fig.9 , shows the function of smart meter in
Fig. 9. Smart meter operation in this scheme
When load demand is high in peak hours, a request is sent
by smart grid to smart meters to reduce the load. In this
case, scheduler takes action and increases the upcoming prices
of next 2 or 3 hours by optimization technique. Therefore,
some portion of load automatically suspends and reduces the
total load. Price predictor and energy scheduler are two main
units to control the residential load. Price predictor estimates
the upcoming prices and allows scheduler to schedule the
appliances according to user’s need.
F. Vickrey-Clarke-Groves (VCG) Mechanism Based DSM
Vickrey-Clarke-Groves (VCG) mechanism in , maxi-
mizes the social welfare i.e. the difference between aggregate
utility function of all users and total energy cost. Authors con-
sider that each user deployed Energy Consumption Controller
(ECC) device in its smart meter for scheduling the household
appliances. ECC schedules the household appliances on run-
time. VCG mechanism develops the DSM programs to enable
efﬁcient energy consumption among all users. In this scheme,
each user provides its energy demand to the utility. By
deploying a centralized mechanism in ECC device, the energy
provider estimates each users optimal energy consumption
level and declares particular electricity payment for each user.
In this way, VCG mechanism reduces the energy cost. Load
demand can be divided into two types i.e. must-run loads and
controllable loads. Must-run loads e.g. a refrigerator that is al-
ways in ON state during the whole day. The controllable loads
are the appliances where operation can be stopped, shifted and
accommodated in different time slots according to demand.
For optimal solution, we evolve an optimization problem to
reduce the total energy cost charged on energy provider while
maximize aggregate utility functions of all users. The solution
of following optimization problem provides an efﬁcient energy
consumption schedule for user’s energy consumption in order
to reduce the cost.
XnPower consumption vector of user n
Un(·)Utility function of user n
Ck(Lk)Cost function of Lkenergy units offered by utility
in each time slot k.
G. A Scheme for Tackling Load Uncertainty
In , an optimization based residential load controlling
algorithm is proposed that tackles the load irregularity to
reduce energy cost in real-time. In algorithm, it is assumed
that each user is facilitated with smart meter. Each smart meter
is equipped with ECC unit. ECC unit schedules and manages
the household energy consumption. In this scenario both real-
time and inclining block rate prices strategies are combined.
Proposed algorithm is formulated as an optimization problem.
Appliances are divided into following categories must run and
controllable loads. Must-run loads start operation immediately
at any time e.g. Personal Computer (PC), TV. These loads start
operation without the interruption of ECC unit. Controllable
appliances operation can be interrupted or delayed.
We separate the operation cycle of appliance into time
slots denoted by T. Each time slot activate with the phase
of admission control. For the starting of appliance operation,
admission control sends an admission request to ECC unit.
Once request is endured, appliance state changes from sleep to
awake. Appliance request acceptance depends on its operation
schedule speciﬁed by ECC unit. ECC unit implements a
centralized algorithm and determines the optimal appliances
schedule in each time slot. This centralized algorithm is
based on an optimization problem formulated for appliances
schedule under different constraints.
TAB L E I
Comparison of different Energy consumption controlling schemes
Scheme Name Method
Scheduling Pricing Coverage
A Model for Autonomous DSM
66.66% NA Real-
time CPP,TOU,RPP Local
Backtracking Based Technique Back-tracking
based scheduling 23.1% NA Real-
time RTP Local
Game Theoretic Model
17% 18% Run
daily load &
ECS Device Based Scheme Energy consump-
tion scheduling 38% 37% Run
daily load &
Optimal Residential Load Control Scheme LP-Based
optimization 22% 10-25% Run
time RTP Local
Vickrey-Clarke Groves Mechanism Scheduling and
Optimization 38% 37.8% Run
Scheme for Tackling the Load Uncertainty
25.5% NA Real
time RTP & I BR Neighbor-hood
In this paper, we have compared different residential load
controlling techniques in the smart grid. Residential load
controlling techniques are employed for efﬁcient consumption
of electricity in residential buildings like homes and ofﬁces.
In current power grid, the load demand curve shows that there
is huge difference between the demand of peak hours and
off-peak hours. The utilities want the load curve to be nearly
smooth to avoid the operation of peaker plants. The load con-
trolling techniques shift some speciﬁc load from peak hours to
off-peak hours and hence helps in making the demand curve
smooth. These techniques reduce the energy consumption cost
and minimize PAR as well as peak load. Consumer should
also be encouraged to schedule the appliances according to
schemes discussed in the paper. Table I shows the comparison
of different schemes. Scheme I reduces the peak load up to
66.66%. So this model is more efﬁcient. ECS device based
scheme and VCG mechanism minimize the cost up to 37%.
 Javaid, Nadeem, et al. “Monitoring and Controlling Power using Zigbee
Communications.” Broadband, Wireless Computing, Communication
and Applications (BWCCA), 2012 Seventh International Conference on.
 Costanzo, Giuseppe Tommaso, et al. “A system architecture for au-
tonomous demand side load management in smart buildings.” (2012):
 Photovoltaics, Dispersed Generation, and Energy Storage. “IEEE Guide
for Smart Grid Interoperability of Energy Technology and Information
Technology Operation with the Electric Power System (EPS), End-Use
Applications, and Loads.” (2011).
 Martins, Rodrigo, and Felipe Meneguzzi. “A Smart Home model to
Demand Side Management.” (2013).
 M. N. Ullah, A. Mahmood, S.Razzaq, M. Ilahi, R.D. Khan, N. Javaid, “A
Survey of Different Residential Energy Consumption Controlling Tech-
niques for Autonomous DSM in Future Smart Grid Communications”,
J. Basic. Appl. Sci. Res., 3(3)1207-1214, (2013).
 Caron, Stphane, and George Kesidis. “Incentive-based energy consump-
tion scheduling algorithms for the smart grid.” Smart Grid Communica-
tions (SmartGridComm), 2010 First IEEE International Conference on.
 Ruiz, Nerea, Iigo Cobelo, and Jos Oyarzabal. “A direct load control
model for virtual power plant management.” Power Systems, IEEE
Transactions on 24.2 (2009): 959-966.
 Wu, Qiuwei, Peng Wang, and Lalit Goel. “Direct load control (DLC)
considering nodal interrupted energy assessment rate (NIEAR) in re-
structured power systems.” Power Systems, IEEE Transactions on 25.3
 Costanzo, Giuseppe T., Jan Kheir, and Guchuan Zhu. “Peak-load shaving
in smart homes via online scheduling.” Industrial Electronics (ISIE),
2011 IEEE International Symposium on. IEEE, (2011).
 Rossello Busquet, Ana, et al. “Reducing Electricity Demand Peaks by
Scheduling Home Appliances Usage.” (2011): 156-163.
 I. Khan, A. Mahmood, N. Javaid, S.Razzaq, R.D. Khan, M. Ilahi, “Home
Energy Management Systems in Future Smart Grids”, J. Basic. Appl.
Sci. Res., 3(3)1224-1231, (2013).
 F. Baig, A. Mahmood, N. Javaid, S. Razzaq, N. Khan, and Z. Saleem,
“Smart home energy management system for monitoring and scheduling
of home appliances using zigbee,” (2013).
 Anas, M., et al. “Minimizing Electricity Theft using Smart Meters in
AMI.” P2P, Parallel, Grid, Cloud and Internet Computing (3PGCIC),
2012 Seventh International Conference on. IEEE, (2012).
 Lee, Junghoon, et al. “Energy consumption scheduler for demand
response systems in the smart grid.” Journal of Information Science
and Engineering 28.5 (2012): 955-969.
 Mohsenian-Rad, A., et al. “Autonomous demand-side management based
on game-theoretic energy consumption scheduling for the future smart
grid.” Smart Grid, IEEE Transactions on 1.3 (2010): 320-331.
 Mohsenian-Rad, A-H., et al. “Optimal and autonomous incentive-based
energy consumption scheduling algorithm for smart grid.” Innovative
Smart Grid Technologies (ISGT), 2010. IEEE, (2010).
 Mohsenian-Rad, A-H., and Alberto Leon-Garcia. “Optimal residential
load control with price prediction in real-time electricity pricing envi-
ronments.” Smart Grid, IEEE Transactions on 1.2 (2010): 120-133.
 Samadi, Pedram, et al. “Advanced demand side management for the fu-
ture smart grid using mechanism design.” Smart Grid, IEEE Transactions
on 3.3 (2012): 1170-1180.
 Samadi, Pedram, et al. “Tackling “the” “Load” Uncertainty Challenges
for Energy Consumption Scheduling in Smart Grid.” (2013): 1-10.