Efficient Resource Utilization in

Cloud-Fog Environment Integrated

with Smart Grids

Presented By:
Junaid Akram
00000119897

Supervised By: GEC Committee:

• Dr. Zeeshan Najam • Dr. Muhammad Tariq Saeed
• Dr. Ammar Mushtaq
Co-Supervised By:
• Dr. Mian Ilyas Ahmad
Outline
• Cloud and Fog
• What is smart grid?
• Problem Statement
• Literature Review
• Motivation
• Objectives
• System Model
• Problem Formulation
• Proposed Approach
• Simulation Setup
• Simulation Results
• Conclusions
Research Center for Modelling and Simulation
1/23/19 2
(RCMS), NUST.
Cloud Computing
• Cloud computing is an umbrella term used to refer to
Internet based development and services.

• A number of characteristics define cloud data, applications

services and infrastructure:
• Remotely hosted: Services or data are hosted on remote
infrastructure.
• Ubiquitous: Services or data are available from anywhere.
• Commodified: The result is a utility computing model similar to
traditional that of traditional utilities, like gas and electricity -
you pay for what you would want!
Research Center for Modelling and Simulation
1/23/19 3
(RCMS), NUST.
Fog Computing
• Intermediate Layer between user and cloud.

• Used for data processing and to serve the immediate requirements of the
end systems.

• Fog nodes can be deployed anywhere with a network connection.

• Any device with computing, storage, and network connectivity can be a fog
node.

• Need for fog:

• Latency-sensitive applications

Research Center for Modelling and Simulation

1/23/19 4
(RCMS), NUST.
Communication and
Energy Networks

Research Center for Modelling and Simulation

1/23/19 5
(RCMS), NUST.
Smart Grid Comparison
with Traditional Grid
Right Now With Smart Grid
Utility doesn’t know when power is used Utilities will offer you lower rates for using
power in “off-peak” times
Utility often relies on you to tell them when Your lights will go out less often and outages
your lights go out won’t last as long
We get large blackouts like the northeast in The grid will automatically create “firebreaks”
2003 fast enough to stop them
Utilities do green power and electric cars as Consumers with green power and electric
“one-offs” cars can be everyday items
Utilities are 10-30 years behind in cyber- Your electric power will not be as vulnerable
security to attackers
Energy prices will increase as aging Prices won’t rise as fast because the system
infrastructure is replaced will be more efficient

Research Center for Modelling and Simulation

1/23/19 6
(RCMS), NUST.
Conceptual View of Smart
Grid

Research Center for Modelling and Simulation

1/23/19 7
(RCMS), NUST.
Cloud Fog and Smart Grid

Source: Zahoor, S., Javaid, S., Javaid, N., Ashraf, M., Ishmanov, F., & Afzal, M. (2018). Cloud–Fog–Based Smart Grid Model for Efficient Resource Management. Sustainability, 10(6), 2079.

Research Center for Modelling and Simulation

1/23/19 8
(RCMS), NUST.
Problem Statement
• In cloud and fog based smart grids, there is an issue regarding
overloading of the requests due to the random arrival of the requests.

• Due to the random utilization of the virtual machines, resources are

sometimes heavily loaded whereas the other resources are idle.

• Load Balancing is a way to distribute the entire load over the network
across a large number of virtual machines.

• This helps to achieve the balanced utilization which maximizes the

performance and minimizes the response time.

Research Center for Modelling and Simulation

1/23/19 9
(RCMS), NUST.
Literature Review
Technique Objective(s) Feature(s) Achievements Limitations
Dragonfly [1] Minimize Schedule the the cost is No
Electricity Cost load demand of minimized and categorization of
home minimize appliances
appliances response time
SSLB [2] Reduce PAR and Allocation of Cost and PAR is Increased
Electricity Cost resources reduced Execution
according to time
user satisfaction
Cloud Load Minimize Utilization of Electricity cost is Cost is not
Balancing [3] Electricity Cost resources reduced balanced
Energy Cost Saving and Confirm Peak to average Fuel emission
Management Load Balancing performance of ratio cells can
System [4] Energy minimization be replaced.

1. Branch, Shahre Rey, and Shahre Rey. "Providing a load balancing method based on dragonfly optimization algorithm for resource allocation in cloud computing." (2018).
2. Li, Changlong, et al. "SSLB: Self-Similarity-Based Load Balancing for Large-Scale Fog Computing.“ Arabian Journal for Science and Engineering(2018): 1-12.
3. Chen, S. L., Chen, Y. Y., & Kuo, S. H. (2017). CLB: A novel load balancing architecture and algorithm for cloud services. Computers & Electrical Engineering, 58, 154-160
4. Varela Souto, A. “Optimization and Energy Management of a Microgrid Based on Frequency Communications.” (2016).

1/23/19 Research Center for Modelling and Simulation (RCMS), NUST. 10

Motivation of Research
• With rapid growth of IoT, Smart Grid has become Computationally intensive.

• Response Time and data storage is critical.

• To manage this Fog and Cloud based architecture is presented.

• Fog Computing acts as an extra level of computational and communication nodes.

• Fog unburden the Cloud from multi-tasking.

• Fog handle large amounts of user requests.

• For the management of user requests different load balancing techniques are defined
in literature.
Research Center for Modelling and Simulation
1/23/19 11
(RCMS), NUST.
Objectives
• Maximum utilization of resources using efficient
load balancing.

• To minimize the Response Time of requests on fog

servers.

• To optimize the Processing Time of user requests on

fog servers.

Research Center for Modelling and Simulation

1/23/19 12
(RCMS), NUST.
Microgrid
• A microgrid is a group of
interconnected loads and
distributed energy
resources.

• A microgrid can connect and

disconnect from the grid.

• This enable it to operate in

both grid-connected or
island mode. https://www.kissclipart.com/micro-grids-clipart-microgrid-renewable-energy-ele-9cetes/download-clipart.html

Research Center for Modelling and Simulation

1/23/19 13
(RCMS), NUST.
System Model

1/23/19 Research Center for Modelling and Simulation (RCMS), NUST. 14

Problem Formulation
• The power load management consists of the
following components:
• set of regions,
• clusters of smart buildings,
• micro grids
• fog servers,
• Cloud
• C be the set of clusters of building
• F be the set of Fogs, F
Research Center for Modelling and Simulation
1/23/19 15
(RCMS), NUST.
Problem Formulation
• The objective function of the proposed research work is to minimize
the processing time and response time.

• Processing time which is to be calculated using the following

parameters:
• Length of the task assigned.
• Capacity of the virtual machines to handle the requests.

• Response time is calculated as difference of finish time and response

time.

• Cost is calculated as the sum of data transfer cost and VM cost.

Research Center for Modelling and Simulation
1/23/19 16
(RCMS), NUST.
Load Balancing
• Load balancing is the process of fair distribution of resources for
efficient resource utilization.

• Resources may include nodes, network resources and processing units.

• Efficient resource utilization means to

• decrease the response,
• increase throughput
• minimizing overheads

• Round Robin is the most common Load balancing Algorithm.

Research Center for Modelling and Simulation

1/23/19 17
(RCMS), NUST.
Round Robin
• Round robin processes the request by selecting the virtual
machines randomly

• The workload gets allocated in the circular manner.

• Round Robin algorithm store entry of user request in the form of a

table.

• Round Robin uses time sharing technique to allocate virtual

machine resources to tasks one after the other to maintain the
balance of multiple requests on virtual machines.
Research Center for Modelling and Simulation
1/23/19 18
(RCMS), NUST.
Odds Algorithm
• The odds algorithm use optimal stopping technique
to search the virtual machine that can available to
perform useful tasks.

• Virtual machines are checked one by one and stop

on the first virtual machine that has enough
capacity to perform tasks.

Research Center for Modelling and Simulation

1/23/19 19
(RCMS), NUST.
Ant Colony Optimization
• ACO algorithm finds the suitable path from large
range by seeking the optimal path.

• In real-world ants roam to find food and after

finding food ants return back to their colony by
laying down a pheromone track.

• The other following ants follow the same path to

find food by searching the shortest path.
Research Center for Modelling and Simulation
1/23/19 20
(RCMS), NUST.
Proposed Approach
• Proposed approach is based on binary particle
swarm optimization with inertia weight adjusted
using simulated annealing.

• Proposed approach is named BPSOSA.

• It is a meta-heuristic self adaptive technique for

load balancing.

Research Center for Modelling and Simulation

1/23/19 21
(RCMS), NUST.
Particle Swarm
Optimization
• PSO is inspired from Bird Flocking and Fish Schooling

• The basic factor in the PSO algorithms is population called as swarm and
solutions called as particles.

• The performance of the algorithms depends upon the local best position and
global best position.

• Each particle contains the special value called as fitness value calculated by
fitness function.

• Objective functions are defined and checked throughout the time by every
particle.
Research Center for Modelling and Simulation
1/23/19 22
(RCMS), NUST.
Binary Particle Swarm
Optimization
• Calculate execution times
• Initialize the swarm
• Set global best

• FOR 0 to NumberOfIterations do
• Calculate inertia weight
• Calculate new velocities
• Calculate new positions
• Calculate fitness value
• Evaluate solution
• Update particle memory
• Update global best
• END FOR

Research Center for Modelling and Simulation

1/23/19 23
(RCMS), NUST.
Simulated Annealing
• Annealing is a thermal process for obtaining low
energy states of a solid in a heat bath.
• The process contains two steps:
• Increase the temperature of the heat bath to a
maximum value at which the solid melts.
• Decrease carefully the temperature of the heat bath
until the particles arrange themselves in the ground
state of the solid. Ground state is a minimum energy
state of the solid.

Research Center for Modelling and Simulation

1/23/19 24
(RCMS), NUST.
Simulated Annealing
• The ground state of the solid is obtained only if the
maximum temperature is high enough and the cooling is
done slowly.

• The search is started with a randomized state.

• We will move to neighboring states always accepting the

moves that decrease the energy while only accepting bad
moves accordingly to a probability distribution dependent
on the “temperature” of the system.
Research Center for Modelling and Simulation
1/23/19 25
(RCMS), NUST.
 Simulation Setup
• Simulation has been run for 24 hours.
• Dynamic service broker policy has been used.
• User bases:
Name Requests Data Peak Peak Average Average
per user Size per hours hours end peak off-peak
per hour Request start (GMT) users users
(bytes) (GMT)
Cluster 1 200 100 11 19 1000 500
Cluster 2 200 100 10 18 1000 500
Cluster 3 200 100 10 17 1000 500
Cluster 4 200 100 10 18 1000 500

Research Center for Modelling and Simulation

1/23/19 26
(RCMS), NUST.
• Fogs Configuration:
Fog Virtual Memory Architectur Operating Virtual
Machines e System Machine
Manager
Fog1 25 512 x86 Linux Xen
Fog2 25 512 x86 Linux Xen
Fog3 30 512 x86 Linux Xen
Fog4 35 512 x86 Linux Xen

Research Center for Modelling and Simulation

1/23/19 27
(RCMS), NUST.
Simulation Results:
Response Time

Research Center for Modelling and Simulation

1/23/19 28
(RCMS), NUST.
Overall Response Time
Summary
• BPSOSA outperforms round robin, odds algorithm
and ant colony optimization by 53.99 ms, 82.08 ms
and 81.58 ms respectively.

Research Center for Modelling and Simulation

1/23/19 29
(RCMS), NUST.
Simulation Results:
Processing Time

Research Center for Modelling and Simulation

1/23/19 30
(RCMS), NUST.
Overall Processing Time
Summary
• BPSOSA outperforms round robin, odds algorithm
and ant colony optimization by 52.94 ms, 81.20 ms
and 80.56 ms respectively.

Research Center for Modelling and Simulation

1/23/19 31
(RCMS), NUST.
Simulation Results: Cost

Research Center for Modelling and Simulation

1/23/19 32
(RCMS), NUST.
Overall Cost Summary
• ACO slightly outperform other algorithms.

Research Center for Modelling and Simulation

1/23/19 33
(RCMS), NUST.
Conclusions
• A system model of cloud and fog base environment integrated with Smart Grid
is presented.

• A novel approach BPSOSA is presented which is based on binary particle swarm

optimization with inertia weight adjusted using simulated annealing.

• In terms of Response Time, BPSOSA outperforms round robin, odds algorithm

and ant colony optimization by 53.99 ms, 82.08 ms and 81.58 ms respectively.

• In terms of Processing Time, BPSOSA outperforms round robin, odds algorithm

and ant colony optimization by 52.94 ms, 81.20 ms and 80.56 ms respectively.

• ACO slightly outperform other algorithms in terms of cost.

Research Center for Modelling and Simulation
1/23/19 34
(RCMS), NUST.
Limitations
• Computationally exhaustive
• No categorization of appliances

Research Center for Modelling and Simulation

1/23/19 35
(RCMS), NUST.
List of Publications
• Junaid Akram, and Zeeshan Najam. “Efficient Resource
Utilization in Cloud-Fog Environment Integrated with
Smart Grids”. In 16th International Conference on
Frontiers of Information Technology (FIT 2018). IEEE,
2018.

• Junaid Akram, and Zeeshan Najam. “Energy Efficient

Localization in Wireless Sensor Networks Using
Computational Intelligence”. In 2018 15th International
Conference on Smart Cities: Improving Quality of Life
Using ICT & IoT (HONET-ICT). IEEE, 2018.
Research Center for Modelling and Simulation
1/23/19 36
(RCMS), NUST.
 Thank You !

Research Center for Modelling and Simulation

1/23/19 37
(RCMS), NUST.