Procedia

Procedia
Computer
Procedia
Computer
Procedia
Science
Computer
Procedia
Available Computer
online Science 00 (2018) 1–9
at www.sciencedirect.com Procedia
Science
Computer
Procedia Computer Science 00 (2018) 1–9
Science
www.elsevier.com/locate/procedia
Computer
ScienceDirect
Procedia Computer Science 00 (2018) 1–9
Science
www.elsevier.com/locate/procedia

Technology Science
www.elsevier.com/locate/procedia
8th International Congress of Procedia Computerand
Information Science 00 (2018) 1–9
Communication (ICICT-2018)
www.elsevier.com/locate/procedia
8th International Congress of Information and Communication Technology (ICICT-2018)
Procedia
Procedia Computer
Computer Science
Science 13100 (2018)
(2018) 1–9
591–599
8th International Congress of Information and Communication Technology
www.elsevier.com/locate/procedia
(ICICT-2018)
A Multi-Objective Quantum-Inspired Genetic Algorithm
8th International Congress of Information and Communication Technology (ICICT-2018)
A Multi-Objective
(Mo-QIGA)
8th International forQuantum-Inspired
Real-Time
Congress of Information Tasks Genetic
Scheduling
and Communication Algorithm
Technology in
(ICICT-2018)
A Multi-Objective
(Mo-QIGA) forQuantum-Inspired
Real-Time Tasks Genetic Algorithm
Scheduling in
A Multi-Objective
(Mo-QIGA) Multiprocessor
forQuantum-Inspired
Real-Time Environment
Tasks Genetic Algorithm
Scheduling in
A Multi-Objective
(Mo-QIGA) Multiprocessor
forQuantum-Inspired
Real-Time Environment
Tasks Genetic Algorithm
Scheduling in
(Mo-QIGA) Multiprocessor
for Real-Time Environment
Tasks b Scheduling in
a,∗
Multiprocessor
Debanjan Konar , Kalpana a
Sharma , Varun Environment
Sarogi , Siddhartha Bhattacharyyac
a,∗
Multiprocessor
Debanjan Konara,∗, Kalpana a
Sharmaa , Varun Environment
a
Sarogibb , Siddhartha Bhattacharyyacc
Debanjan Konar , KalpanaMajitar, SharmaSikkim,
Sikkim Manipal
, Varun SarogiIndia, Siddhartha Bhattacharyya
Institute of Technology,
a SikkimEast Sikkim-737136,
Debanjan Konara,∗, Kalpana Sharma
b ThoughtWorks
Majitar,
a Institute
Manipal
, Varun
Technologies,
a SikkimEast Sikkim,
Manipal
Sarogi
Haryana-122002,
Sikkim-737136,
Institute of
b
of Technology,
, Siddhartha
India
Technology,
India Bhattacharyyac
Debanjan Konara,∗, ofKalpana
c RCC Institute Information Sharma
b ThoughtWorks
Majitar,
a
, Varun
Technology,Canal South
Technologies,
a SikkimEast Sikkim,
Manipal
Sarogi
Road,
Sikkim-737136,
Institute
b
, Siddhartha
Beliaghata,
Haryana-122002,
India
of Technology,
Bhattacharyya
Kolkata - 700015,
India India c
c RCC Institute of Information Technology,Canal
b ThoughtWorks South Road, Beliaghata, Kolkata - 700015, India
Majitar, Technologies, Haryana-122002,
India India
c RCC a SikkimEast Sikkim,
Manipal Sikkim-737136,
Institute of Technology,
Institute of Information Technology,Canal
b ThoughtWorks South
Technologies, Road, Beliaghata,
Haryana-122002, Kolkata - 700015, India
Majitar, East Sikkim, Sikkim-737136, India India
c RCC Institute of Information Technology,Canal South Road, Beliaghata, Kolkata - 700015, India
b ThoughtWorks Technologies, Haryana-122002, India
c RCC Institute of Information Technology,Canal South Road, Beliaghata, Kolkata - 700015, India
Abstract
Abstract
This paper aims at providing a multi-objective real-time scheduling algorithm suitable for scheduling of real-time tasks
Abstract
in multiprocessor
This paper aims atsystem witha multi-objective
providing multi-objective criteria.
real-timeInscheduling
this proposed Mo-QIGA,
algorithm scheduling
suitable of multi-objective
for scheduling real-
of real-time tasks
Abstract
time
in tasks have
multiprocessorbeen targeted
system with
This paper aims at providing a for minimizing
multi-objectivecompletion
criteria. Intime
this and total
proposed tardiness
Mo-QIGA, of each real-time
scheduling of task simultaneously.
multi-objective
real-time scheduling algorithm suitable for scheduling of real-time tasksreal-
Abstract
The
time
in exploration
tasks have
multiprocessorofsystem
quantum
been computation
targeted
with for in the
minimizing
multi-objective suggested
completion
criteria. In Mo-QIGA
time
this and mimics
total
proposed the principles
tardiness
Mo-QIGA, of each and basic
real-time
scheduling of concepts
task of quan-
simultaneously.
multi-objective
This paper aims at providing a multi-objective real-time scheduling algorithm suitable for scheduling of real-time tasks real-
tum
The
time mechanics.
exploration
tasks have Variable
of quantum length chromosomes
computation in the are used in
suggested Mo-QIGA
Mo-QIGA and
mimics inthe
order to exhibit
principles andfull
basicHilbert
conceptshyperspace,
of quan-
in
qubitspaper aimsbeen
multiprocessor
This representation
targeted
atsystem
providing
is
with afor
preferred.
minimizing
multi-objective
multi-objective
In addition,
completion
criteria.
real-time
Quantum
time
this and
Inscheduling
rotation
total
proposed tardiness
Mo-QIGA,
algorithm
gate is
of each
suitable
employed to forreal-time
scheduling
update scheduling
the
taskofsimultaneously.
of multi-objective
schedules real-time real-
tasks
andofit quan-
obvi-
tum exploration
The
time mechanics.
tasks have Variable
of
beenquantum length
targeted for chromosomes
computation
minimizingin the are used in
suggested
completion timeMo-QIGA
Mo-QIGA
and and
mimics
total inthe
tardiness order to exhibit
principles
of each and full
real-timebasicHilbert
taskconceptshyperspace,
simultaneously.
in multiprocessor
ates classical system
genetic with multi-objective
operators. Inchromosomes
order criteria.
to obtain In this proposed Mo-QIGA, scheduling of multi-objective real-
qubits
tum representation
mechanics.
The exploration
time tasks have of
been
is preferred.
Variable
quantum length
targeted
In addition,
computation
for minimizingin the arePareto-optimal
Quantumused rotation
suggested
completion in Mo-QIGA
Mo-QIGA
time
solutions,
gatemimics
and total
isand ainrandom
employedorder
the
tardiness
totokey
updatedistribution
eachexhibit
principles
ofbased
the
and full
real-timebasic
has been
schedules
Hilbert
taskconceptsandofadopted
it quan-
obvi-
hyperspace,
simultaneously.
to convert
ates the
classical qubits
genetic particles
operators. to valid schedule
Inchromosomes
order to obtain solutions. Moreover, permutation trimming technique has been
qubits representation
tum exploration
The mechanics. is preferred.
Variable
of quantum length In addition,
computation arePareto-optimal
in theQuantumused rotation
suggested inMo-QIGA
Mo-QIGAsolutions,
gatemimics
isand ainrandom
employedorder
the totokey
updatedistribution
exhibit
principles the
and full
basic
has been
schedules
Hilbert
conceptsandofadopted
it quan-
obvi-
hyperspace,
introduced
to convert
ates classicalforgenetic
the population
qubits diversification
particles
operators. toInvalid
order which
schedule
to leads
obtain to good
solutions. quality
Moreover,
Pareto-optimal schedules.
permutation
solutions, a The
random experiments
based
key trimming have
distribution been
technique
has carried
been has out
been
adopted
qubits representation
tum mechanics. is
Variable preferred.
length In addition,
chromosomes Quantum
are used rotation
in Mo-QIGAgate is employed
and in order to update
to exhibit the schedules
full Hilbert and it obvi-
hyperspace,
using
to number
introduced
convert for
the of processors
population
qubits particles varying
to with
diversification
valid real-time
which
schedule leads tasks
to
solutions. priority
good andschedules.
quality
Moreover, processing
permutation time.
The It trimming
has been
experiments
based found
have been that
technique the been
carried
has sug-
out
ates classical
qubits genetic operators.
representation isquantum
preferred. In order to obtain
In addition, Pareto-optimal
Quantum rotation solutions,
gate is its a random
employed key distribution
tocounterpart
update the in has been
schedules and adopted
it obvi-
gested
using multi-objective
number
introduced for of processors
population inspired
varying with
diversification genetic
which algorithm
real-time
leads tasks
to good outperforms
priority and
quality classical
processing
schedules. time.
The It has
experiments been terms
found
have beenofthat
scheduling
the
carried sug-
out
to convert
ates the qubits operators.
classical particles toInvalid schedule
to obtainsolutions. Moreover, permutation based
key trimming technique
has beenhas been
accuracy
gestednumber
using and genetic
time
multi-objective
of in multi-objective
quantum
processors varying
order
sense.
inspired
with genetic
real-time
Pareto-optimal
algorithm
tasks priority
solutions,
outperforms
and
a random
its classical
processing time. It
distribution
counterpart
has inhave
been terms
found ofthat
adopted
scheduling
the sug-
introduced for population diversification which leads to good quality schedules. The experiments
to convert the qubits particles to valid schedule solutions. Moreover, permutation based trimming technique has been been carried out
accuracy
gested
using and time
multi-objective
number of in multi-objective
quantum
processors varying sense.
inspired
with genetic algorithm
real-time tasks outperforms
priority and its classical
processing counterpart
time. It has in
been terms
found of scheduling
that the sug-
© 2018 Thefor
introduced Authors. Published
population by Elsevier
diversification Ltd.leads to good quality schedules. The experiments have been carried out
which
c 2018
accuracy The time
Authors. Published bysense.
Elsevier B.V. Peer-review under itsresponsibility of organizing committee of the
anand in multi-objective

gested multi-objective quantum inspired genetic algorithm outperforms classical counterpart in terms
This
using isnumber
openofaccess article
processors under
varying the CC
with BY-NC-ND
real-time taskslicense
priority foundofthat
(https://creativecommons.org/licenses/by-nc-nd/4.0/)
and processing time. It has been scheduling
the sug-
8th
c International
2018
Selection The Congress
Authors.
and peer-review of Information
Published by
underinspired and
Elsevier
responsibility Communication
B.V. Peer-review Technology
under (ICICT-2018).
responsibility of organizing committee of the
accuracy
 and time
gested multi-objective in multi-objective
quantum geneticofalgorithm
sense. the scientific committee
outperforms itsof the 8th counterpart
classical InternationalinCongress of Information
terms of scheduling and
8th
c International
2018 The
Communication

accuracy Congress
Authors.
Technology.
Keywords:andMultiprocessor
of Information
Published
time in multi-objective by and
ElsevierCommunication
B.V. Peer-review Technology
under (ICICT-2018).
responsibility of
sense. task, task scheduling, quantum inspired genetic algorithm (QIGA)
system, real-time
organizing committee of the
8th
 International
c 2018 Congress
The Authors. of Information
Published and Communication
by Elsevier B.V. Peer-review Technology (ICICT-2018). of organizing committee of the
under responsibility
Keywords: Multiprocessor system, real-time task, task scheduling, quantum inspired genetic algorithm (QIGA)
8th
c International
2018 The Congress
Authors. of Information
Published by and Communication
Elsevier B.V. Peer-review Technology
under
Keywords: Multiprocessor system, real-time task, task scheduling, quantum inspired genetic
(ICICT-2018). of organizing committee of the
responsibility algorithm (QIGA)
1.
8th Introduction
International Congress of Information and Communication Technology (ICICT-2018).
Keywords: Multiprocessor system, real-time task, task scheduling, quantum inspired genetic algorithm (QIGA)
1. Introduction
Real-time
Keywords:
1. schedulingsystem,
Multiprocessor
Introduction in multiprocessor
real-time task, environment
task scheduling, is anquantum
uphill task and genetic
inspired a rigorous studied
algorithm problem in
(QIGA)
the field of computer
Real-time
1. Introduction scheduling science and engineering
in multiprocessor [1, 2, 3, 4].
environment is anThe various
uphill applications
task and a rigorousofstudied
real-time systems
problem in
include
the fieldmedical
Real-time
of imaging,
scheduling
computer space
in
science system,
and process
multiprocessor
engineering control,
environment
[1, 2, 3, flight
is
4].an control,
uphill
The sensor
task
various and a networks,
rigorous
applications ofrobotics,
studied satellites,
problem
real-time systems in
1. Introduction
scientific
the field
include experiments,
of computer
medical
Real-time unmanned
science
imaging, in
scheduling spaceand vehicles etc.
engineering
system, process
multiprocessor Real-time
[1, 2,
environment 3, tasks
4]. The
control, flight are classified
various
control,
is an uphill as hard
applications
sensor
task and
and anetworks,ofsoft real-time
real-time tasks
systems
robotics,problem
rigorous studied satellites,
in
relying
the fieldon
include
scientific ofthe
medical
Real-time deadline
imaging,
experiments,
computer
scheduling compliance
space
unmanned
science
in and [5, 6,process
system,
vehicles 7,
engineering
multiprocessor 8].Real-time
etc. The
[1, 2,principle
control,
environment flight
tasks
3, 4].
is characteristic
control,
anThe sensor
arevarious
classified
uphill task ofanetworks,
as hard
hard
applications
and real-time
and
ofsoft
rigorous systems
robotics,
real-time
real-time
studied is to
satellites,
tasks
systems
problem in
scientific
relying
fieldon
include
the experiments,
the
medical
of deadline
imaging,
computer unmanned
compliance
spaceand
science vehicles
system, etc.
engineering 8].Real-time
[5, 6,process
7, The
control,
[1, tasks
2,principle
flight
3, 4]. arevarious
classified
characteristic
control,
The sensoras hard
hard and
ofnetworks,
applications ofsoft
real-timereal-time
systems
robotics,
real-time tasks
is to
satellites,
systems
relying
scientific
include on the deadline
experiments,
medical imaging, compliance
unmanned [5, 6,process
vehicles
space system, 7, 8].Real-time
etc. The principle
control, tasks characteristic
flight are classified
control, ofnetworks,
sensoras hard and
hard real-time
soft systems
real-time
robotics, is to
tasks
satellites,
∗ Corresponding author. Tel: +91-9821928536
relying onexperiments,
scientific the deadlineunmanned
compliance [5, 6, 7,
vehicles 8].Real-time
etc. The principle
tasks characteristic of hard
are classified as hard and
real-time systemstasks
soft real-time is to
∗ Email address: author.
Corresponding [email protected]
Tel: +91-9821928536 (Debanjan Konar)
relying on the deadline compliance [5, 6, 7, 8]. The principle characteristic of hard real-time systems is to
∗ Email address: author.
Corresponding [email protected]
Tel: +91-9821928536 (Debanjan Konar)
© 2018∗ Email
The Authors. Published
address: by Elsevier B.V. This is an open access(Debanjan
[email protected] article underKonar)
the CC BY-NC-ND license
Corresponding author. Tel: +91-9821928536
1877-0509
The © 2018 The
Authors. Authors.
Published
address:
Published by Elsevier Ltd.
https://creativecommons.org/licenses/by-nc-nd/4.0/)
© 2018∗ Email by Elsevier B.V. This is an open access(Debanjan
[email protected] article underKonar)
the CC BY-NC-ND license
This is Corresponding
an open accessauthor.
article Tel:
under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)
+91-9821928536
© 2018 Email
SelectionThe
andAuthors.
and Published
peer-review
peer-review
address: by Elsevier
underunder B.V.
responsibility Thisscientific
of the
https://creativecommons.org/licenses/by-nc-nd/4.0/)
Selection responsibility
[email protected] isofanthe
open access(Debanjan
committeearticle
scientific theunder
ofcommittee the CC
8th Konar) BY-NC-ND
International
of the license
Congress
8th of Information
International and Communication
Congress of Information and
Technology
https://creativecommons.org/licenses/by-nc-nd/4.0/)
© 2018 The andAuthors.
Communication
Selection Published
under by
Technology
peer-review Elsevier B.V.
responsibility Thisscientific
of the is an open access article
committee of theunder the CC BY-NC-ND
8th International Congresslicense
of Information and Communication
10.1016/j.procs.2018.04.301
Selection andAuthors.
peer-review under by
responsibility of the
Technology
https://creativecommons.org/licenses/by-nc-nd/4.0/)
© 2018 The Published Elsevier B.V. Thisscientific
is an opencommittee of theunder
access article 8th International Congresslicense
the CC BY-NC-ND of Information and Communication
Technology
Selection and peer-review under responsibility of the scientific committee of the 8th International Congress of Information and Communication
https://creativecommons.org/licenses/by-nc-nd/4.0/)
Technology
Selection and peer-review under responsibility of the scientific committee of the 8th International Congress of Information and Communication
Technology
592 Debanjan Konar et al. / Procedia Computer Science 131 (2018) 591–599
2 D.Konar et al. / Procedia Computer Science 00 (2018) 1–9

ensure the total deadline compliance and failure to meet deadlines leads to catastrophic consequences. On
the other hand, the performance measure of soft real-time systems lies in the optimization of deadline com-
pliance [7, 9]. Generally, processor utilization, throughput, number of tasks missing the deadlines etc. are
the key parameters of real-time task scheduling algorithms. In uniprocessor system, there are rate monotonic
(RM) and earliest deadline first (EDF) heuristics which have been used to derive scheduling algorithms for
a hard real-time systems [10, 11]. The important aspect of deadlines compliances for hard and soft real-time
tasks are assessed by various scheduling techniques. However, these techniques provide optimal solution in
some restricted environments. It also suffers in coping up with soft real-time tasks pertaining to utilization
of resources and not suitable for high stake tasks.
The primary focus of soft real-time task scheduling lies in minimizing total tardiness of each task. Recently,
Kim et. al proposed a rate regulating proportional share (RRPS) scheduling algorithm relying on stride
scheduler [12]. A modification on proportional share (mPS) scheduling has been done for soft real-time task
scheduling and it is reported as efficient scheduling algorithm in uniprocessor system over other traditional
scheduling approaches. In last few decades, the wide variety in scheduling criteria of multi-processor real-
time task scheduling problems have been evolved using various research initiatives based on several heuris-
tics and meta heuristics [13, 14], simulated annealing, evolutionary algorithms [15, 16, 17, 18, 19, 20, 21, 22]
and hybrid techniques [23, 24, 25]. Considering the real-time tasks as non-preemptive with precedence
and deadlines, H. Mitra et al. [26] suggested an efficient algorithm for real-time task scheduling and also
compared with the minimum-laxity-first heuristic. To schedule partially ordered non-preemptive tasks in
multiprocessor environment and to minimize the completion time of each task, different time varying oper-
ators have been employed by Lin et al. [27] with the lifetime of each task. However, the efficiency of the
proposed approach is not convincing compared to Tabu search. In addition, Monnier et al. [28] proposed a
non-preemptive genetic algorithm for real-time task scheduling with deadline constraint.
Due to enormity in complexity and scale, traditional real-time systems are not suitable to meet the present
demand of high performance computing. A plethora of multiprocessor real-time tasks scheduling concern
with the optimization of single parameter make-span or the number of scheduled tasks missing deadlines. In
practice, to improve efficiency of the overall real-time systems, multi-objective real-time tasks scheduling
is suitable considering all aspects and trade-offs in scheduling problems [15, 16, 17, 18]. Recently, Wu
et al. [29] presented a multiprocessor scheduling scheme with non-preemptive soft real-time tasks using
multi-objective genetic algorithm. The multi-objective criteria are to minimize the total tardiness and the to-
tal number of processors used. In addition, multi-objective real-time tasks scheduling is a NP hard problem
and the tasks of researchers become more challenging than scheduling problems with single criterion. How-
ever, the main problem of scheduling of multi-objective real-time tasks in a multiprocessor system are higher
turnaround time due to inconsistent and conflicting behavior of multi-objective criteria [15, 16, 17, 18]. Re-
cent years have witnessed enormous application of classical Genetic Algorithms (GA) approaches in the
area of multi-objective real-time scheduling in multiprocessor environment [15, 16, 30, 17]. Generally,
optimal scheduling of real-time tasks in multiprocessor environment to minimize tardiness is a NP complete
problem [31]. Genetic algorithm exhibits new population and produced better average fitness value of the
new population. Selection of fitness function, population size, crossover probability and mutation probabil-
ity are key aspects behind the improvement in the average fitness value. In real-time scenario, the problem
under consideration is trade offs between two objectives which are conflicting in nature referred to as Pareto
optimum. However, the aforementioned genetic algorithm based approaches do not consider the conflict
between objectives and moreover, the inherent problem lies with the genetic algorithm based approaches is
lack of producing valid schedules owing to crossover and mutation operation.
Quantum computing is an emerging field of engineering depending on quantum mechanical effects to
solve computational problems [32]. Using such unique quantum mechanical phenomena, various optimiza-
tion problems can be solved efficiently. On the contrary to some novel quantum algorithms [33, 34] consid-
ered, these algorithms can be simulated on classical computers with great precision. However, it is extremely
computationally exhaustive. Nevertheless, inherent randomness offered by representation of qubits is drawn
from the basic properties and concepts of quantum mechanics. First evolutionary algorithm inspired by
quantum computing was proposed by A. Narayan et al. [35] in 1996 and this field is still vigorously studied
nowadays. The combination of evolutionary algorithms and quantum computing gains popularity due to
 Debanjan Konar et al. / Procedia Computer Science 131 (2018) 591–599 593
D.Konar et al. / Procedia Computer Science 00 (2018) 1–9 3

the inherent probabilistic behavior of quantum computing which leads to good exploration of global search
space. The notable contribution includes in this Quantum Inspired Evolutionary Computing (QEA) is by
Han et. al [36, 37, 38, 39, 40, 41]. They proposed a Quantum Inspired Genetic Algorithm (QIGA) suitable
for various single and multi-objective scheduling. Recently, Konar et.al [42, 43] also proposed an efficient
Hybrid Quantum Inspired Genetic Algorithm (HQIGA) appropriate for Real-time task scheduling in multi-
processor environment. It has been reported that HQIGA outperforms its classical counterparts in terms of
time and accuracy.
In this current work, a dynamic multi-objective real-time tasks scheduling is proposed in multi-processor
environment using Multi-objective Quantum Inspired genetic algorithm (MoQIGA) guided by multi-objective
fitness functions. The proposed Mo-QIGA algorithm aims to solve multi-objective scheduling by minimiz-
ing total tardiness and completion time of the non-preemptive real-time tasks with precedence relationship
concurrently. In MoQIGA, constituent bits of chromosomes with varying length are characterized as prob-
abilistic representation of qubits. Owing to inherent probabilistic nature of qubits, quantum chromosomes
have better population diversity in schedules. In this proposed work, total tardiness and completion time of
the tasks in multi-objective sense are used to govern the fitness value for each chromosome individual. A
quantum measurement is applied on each individual schedule, obtained using rotation gate operation.

2. Multi-Objective Real-Time Task Scheduling Problem and Task Model

In multi-objective real-time systems, the real-time tasks with precedence are represented as task graph
associated with deadlines. Assuming all real-time tasks as non-preemptive, the primary focus of the multi-
objective scheduling in multi-processor environment is to minimize total tardiness (S 1 ) and completion time
(S 2 ). Formally, aforementioned multi-objective real-time task scheduling problem can be formulated with
m number of tasks in real-time environment and n number of identical processors as follows.

min g(x) = (g1 (x), . . . , gm (x))∀x ∈ X (1)

where g1 , . . . , gm are the different objective functions in distributed environment, x is the variables of vec-
tors and the valid set of schedule solutions are denoted by X. If a non-dominated or Pareto optimal solution
schedule x∗ exists, then for all x ∈ X, x ≺ x∗. However, problem under consideration, is subset of a larger
number of objective function.
A real-time system in multi-processor system comprises with m number of real time tasks T = {T 1 , T 2 , T 3 , . . . T m }
and n identical processors P = {P1 , P2 , P3 , . . . Pn }. Each task T i ∈ T is characterized [16] by Ai (S ) : arrival
time of the task T i in solution S ; Ri (S ): ready time of the task T i in solution S ; Ci (S ) : worst case computa-
tion time of the task T i in solution S ; Di (S ) : deadline of the task T i in solution S ; Bi (S ) : earliest start time
of the task T i in solution S ; Fi (S ) : completion time of the task T i in solution S . The task T i is said to meet
its deadlines if Ri (S )  Bi (S )  Di (S ) − Ci (S ) and Ri (S ) + Ci (S )  Fi (S )  Di (S ) [6].

Assuming all real-time tasks are non-preemptive, Fi = Bi + m k=1 pik . The laziness of a task T i in solution
S can be defined as Li (S ) = Fi (S ) − Di (S ) and the tardiness of the task T i can be represented as T Ri (S ) =
max{Li (S ), 0}. In addition to this, earliness of the task T i also can be defined as ERi (S ) = max{−Li (S ), 0}.
The maximum completion or finish time or make span of task T i : Fmax(S ) = maxi {Fi (S )}. Owing to multi-
objective scheduling criteria in real-time, each task T i is assigned with a weight Wi . Due to the conflicts
among the different objectives in multi-objective scheduling, the primary focus of this project is to explore
efficient and effective solutions satisfying all objectives in real-time. Moreover, it is always desirable to
obtain schedule solutions with good proximity and diversified solutions.
The proposed multi-objective scheduling in multiprocessor environment is formulated to minimize total tar-
diness of and completion time of the real-time tasks and it can be modeled [16] as follows:
The total tardiness of a tasks in solution S is minimized as
m
 n

min S 1 = max{0, (Bi (S ) + Ci j (S ) − Di (S ))αi j } (2)
i=1 j=1
594 Debanjan Konar et al. / Procedia Computer Science 131 (2018) 591–599
4 D.Konar et al. / Procedia Computer Science 00 (2018) 1–9

The total completion time of real-time tasks in Solution S is also

min S 2 = max{Fi (S )} (3)

provided
Bi (S )  Fi (S ) ∀i (4)
n
Bi (S )  Bk (S ) + αk j , T k = pred(T i ) ∀i (5)
j=1
n

αi j = 1 ∀i (6)
j=1

αi j ∈ {0, 1} ∀i, j (7)

The set of immediate predecessors of real-time task T i is defined as pred(T i ). The finish time of each real
time task T i in solution S is defined as

Fi (S ) = min{Bi (S ) + Ci(S ), Di (S )} (8)

If the processor p j is assigned for the task T i then αi j is set to 1, other wise it is set to 0.

3. Quantum-Inspired Genetic Algorithm (QIGA)

Quantum Inspired Genetic Algorithm (QIGA) is a quantum version of classical genetic algorithm, which
exploits the power of quantum computation in order to obviate the usage of classical genetic operations like
selection, crossover and mutation. The concept and principles of the quantum computation are motivated
by quantum mechanics. Researchers in recent time witnessed the QIGA proposed by Han et al. [36, 37]
and it gained popularity due to novel probabilistic representation of individual chromosome in the form of
qubits (Q bit). The probabilistic representation of chromosome has better diversified population character-
istics than classical binary representation. The basic necessary concepts pertaining to quantum computing
are briefly explained as follows:

3.1. Concept of Qubits

In quantum computing basic building block is a quantum bit or qubit [44] for processing of information.
The linear superposition of dual eigenstates |0 and |1 constitutes a qubit in quantum computer. It is defined
as
|ρ = α|0 + β|1 (9)
The probabilities for occurrence of |0 and |1 eigenstates are |α|2 and |β|2 respectively where, α and β are
complex numbers and subjected to normalization constraint

|α|2 + |β|2 = 1 (10)

3.2. Single qubit rotation gate

The quantum logic gates are implemented on Hilbert space using various linear and unitary operations.
Updation of single qubit is done using a rotation gate is as follows
      
α cos φ − sin φ α
= × (11)
β sin φ cos φ β
A qubit (α, β) is modified to (α , β ) using single qubit rotation gate with an angle φ.
 Debanjan
D.Konar Konar etComputer
et al. / Procedia al. / Procedia Computer
Science Science
00 (2018) 1–9 131 (2018) 591–599 5 595

Algorithm 1: Quantum-Inspired Genetic Algorithm

1 Initialize τ → 0
2 An initial population

QP(τ) = {QC1 (τ), QC2 (τ), QC3 (τ), QC4 (τ), . . . , QCm (τ)}

is explored randomly. The kth quantum behaved individual chromosome QCk (τ) constituted with
qubits, generating in τth iteration as
 
α1 (τ) α2 (τ) α3 (τ) . . . αn (τ)
QCk (τ) =
β1 (τ) β2 (τ) β3 (τ) . . . βn (τ)

3 Apply quantum measurement on the initial population of quantum behaved chromosomes QP(τ) to
produce true solutions CS (τ) and all solutions available in it are evaluated
4 On evaluating the best solution from all true solutions CS (τ) available, it is stored in λ(τ)
5 Preserve the best solution in λ(τ) on satisfying the terminating condition, otherwise
6 Do
7 τ→τ+1
8 To obtain new population QP(τ) from old version QP(τ + 1), rotation gate is employed with a
rotation angle ω as  
cos ω − sin ω
PQ(τ) = QP(τ − 1)
sin ω cos ω
9 Apply quantum measurement on the initial population of quantum behaved chromosomes
QP(τ + 1) to obtain true solutions CS (τ + 1) and all solutions available in it are evaluated
10 The best solution among CS (τ + 1) and λ(τ) are accumulated into the storage λ(τ + 1)
11 Till the terminating condition is false

Due to the inherent probabilistic and random behavior of constituent qubits in quantum chromosomes,
QIGA establishes a new dimension in to the proposed algorithm. A string of qubits or quantum bits consti-
tutes each chromosome individual, QC can be represented as a string of n qubits which is encoded as
 
α 1 α 2 α 3 . . . αn
QC = (12)
β1 β2 β3 . . . βn
where
|αi |2 + |βi |2 = 1, i = 1, 2, 3, ...n. (13)
The structure of quantum behaved chromosome is illustrated in Figure 1 with quantum genes.

Fig. 1. Structure of quantum behaved chromosome [42, 43]

The QIGA algorithm [42, 43] is explained as follows.

4. Proposed Multi-objective Quantum-Inspired Genetic Algorithm (Mo-QIGA)

The Mo-QIGA, which mimics the QIGA procedure for real-time tasks with multi-objective criteria
scheduling in multiprocessor environment. The Mo-QIGA algorithm is illustrated as below.
596
6 Debanjan
D.Konar Konar et Computer
et al. / Procedia al. / Procedia Computer
Science Science
00 (2018) 1–9 131 (2018) 591–599

Algorithm 2: Multi-objective Quantum-Inspired Genetic Algorithm

1 Build the task graph of size M with priority
2 Do
3 Create the ready queue using the tasks arrived for execution
4 If the ready queue size  1 then
5 Set the size of chromosome (CS) according to the size of the ready queue
6 Randomly produce initial chromosomes (population) of size N
7 Evaluate each gene of the quantum behaved chromosomes (valid Population of real-time tasks)
using quantum observation and allocate to processors
8 Select the valid schedule solution (population) considering the suggested multi-objective fitness
function
9 Execute the fast non-dominated sort and crowding distance sort of their fitness values
10 Select the superior chromosome among the population
11 Do
12 Produce new population applying quantum rotation gates on qubits constituted chromosomes with
the superior chromosome
13 Apply quantum observation on new population to obtain binary chromosomes and evaluate the
multi-objective fitness value of each chromosome
14 Execute the fast non-dominated sort and crowding distance sort of their fitness values
15 Select the superior chromosome among the population and schedule the task in the corresponding
processor
16 Until the superior chromosome satisfy the scheduling criteria
17 Until size of the execution queue != M

Quantum chromosomes are constituted using qubits. The multi-criteria fitness values of each quantum
behaved chromosomes can not be determined owing to the probabilistic nature of qubits. In order to evaluate
it and to obtain binary form of the individual chromosomes, a quantum observation [37] is applied on the
qubits constituted chromosomes. A look up table has been introduced to tune the rotation angle pertaining
to rotation gate.
The quantum observation followed by random key distribution [45, 46] allows the quantum behaved
qubits constituted chromosomes to convert into binary chromosome in the suggested Mo-QIGA procedure.

5. Results and Analysis

The current work considers 100, 200 and 300 randomly produced real-time tasks and these are allocated
to the identical processors (variable size chromosomes) of sizes 10, 15, 20, 25, 30 with population size (4∗ np )
where n and p are the number of real-time tasks and processors respectively. Multi-objective fitness func-
tion has been employed to find the fitness value of each task. The non-dominated sorting and crowding
distance sorting methods are applied on the fitness values of the real-time tasks. The proposed Mo-QIGA
and Multi-objective Classical Genetic Algorithm (Mo-CGA) have been simulated in C, which provides ro-
bust available platform to implement real-time scheduling in multi-processor environment along with all
type of complexities due to multi-objective criteria. In Mo-CGA, the crossover probability is set to 0.5 and
mutation probability ranges from 0.1 to 0.2 with varying population size of (4 ∗ np ). Table 1, Table 2 and
Table 3 are reported to show the efficacy of the suggested Mo-QIGA over classical counter part calculating
the percentage of success (%) [43] as follow.
total number o f scheduled real − time tasks
% o f success = ∗ 100% (14)
total number o f real − time tasks
 Debanjan
D.Konar Konar etComputer
et al. / Procedia al. / Procedia Computer
Science Science
00 (2018) 1–9 131 (2018) 591–599 7 597

Algorithm 3: Quantum Observation

1 k→0
2 Do
3 k =k+1
4 If rnd[0, 1) < |αk |2
5 Return 0
6 Else
7 Return 1
8 End if
9 Until k  n

Table 1. Percentage of success obtained for 100 number of tasks

Mo-QIGA Mo-CGA
No. of % of Computation % of Computation
Processos success Time success Time
(P) (ms) (ms)
10 94.764 3.250385 93.744 10.08163
15 94.000 2.060282 91.048 9.80149
20 94.618 2.170402 93.300 8.44273
25 94.834 1.790288 93.534 7.71716
30 90.452 1.768280 89.390 8.28760

It may be noted from the above results that an increase the number of processors with the same number
of real-time tasks causes a percentage drop in efficiency of the proposed algorithm. In addition, the number
of tasks and the number of processors are directly proportional to the time required for scheduling.

6. Conclusions

A novel real-time task scheduling using multi-objective quantum inspired genetic algorithm (Mo-QIGA)
has been presented in this paper. However, none of the existing algorithms do not offer guarantee of opti-
mality. Mo-QIGA also falls in the same category. The primary objective of the suggested algorithm is to
minimize total completion time and tardiness of the real-time tasks. Variable length chromosomes are used
in Mo-QIGA and in order to exhibit full Hilbert hyperspace, qubits representation is preferred. In addition,
Quantum rotation gate is employed to update the schedules and it obviates classical genetic operators.

Table 2. Percentage of success obtained for 200 number of tasks

Mo-QIGA Mo-CGA
No. of % of Computation % of Computation
Processos success Time success Time
(P) (ms) (ms)
10 94.132 12.71182 91.894 33.89512
15 93.086 8.23604 90.753 23.28135
20 94.076 6.49097 91.188 22.37334
25 94.179 5.35073 92.182 28.38435
30 92.773 5.06070 91.666 42.93649
598
8 Debanjan
D.Konar Konar et Computer
et al. / Procedia al. / Procedia Computer
Science Science
00 (2018) 1–9 131 (2018) 591–599

Table 3. Percentage of success obtained for 300 number of tasks

Mo-QIGA Mo-CGA
No. of % of Computation % of Computation
Processos success Time success Time
(P) (ms) (ms)
10 94.135 25.29012 90.888 65.26607
15 93.526 16.85241 91.099 43.44666
20 93.077 13.54205 91.379 38.28576
25 93.070 11.64360 91.320 44.37106
30 93.214 10.10153 91.345 55.96840

The algorithm proposed here has been successful in obtaining feasible solutions for a task set between
100 to 300 and processors between 10 to 30 with almost 92% and also achieving high utilization of pro-
cessors. It is noted that the implementation of the Mo-QIGA is quite costly since populations of solutions
are coupled with computation intensive fitness evaluations. This can be overcome by employing distributed
parallel processing technique in multi-processor computing domain. There are different methods remain
to investigate the application of preemptive real-time tasks in distributed environment. The authors are
currently focused in this direction.

References
[1] S. C. Cheng, D. F. Shiau, Y. M. Huang and Y .T. Lin, “Dynamic hard-real-time scheduling using genetic algorithm for multipro-
cessor task with resource and timing constraints”, Expert Systems with Applications. Vol. 36, pp. 852–860, 2009.
[2] G. Manimaran and C. Siva Ram Murthy, “An Efficient Dynamic Scheduling Algorithm for Multiprocessor Real-time Sys-
tems”,IEEE Transactions on Parallel and Distributed Systems, Vol. 9, No.3 pp. 312–319, 1998.
[3] E. Seo, J. Jeong, S. Park and Joonwon Lee,“Energy Efficient Scheduling of Real-Time Tasks on Multicore Processors”, IEEE
Transactions on Parallel and Distributed Systems. Vol. 19, No. 11 pp. 1540–1552, 2008.
[4] A. Srinivasan, J. H. Anderson, “Fair Scheduling of Dynamic Task Systems on Multiprocessors”, The Journal of Systems and
Software, Vol. 77, pp. 67–80, 2005.
[5] L. Budin, J. Domagoj and G. Marin, “Genetic algorithm in real-time imprecise computing”, Proceedings of Industrial Electronics,
pp. 84-89.
[6] E. Eggers. 1999, “Dynamic Scheduling Algorithms in Real-time, Multiprocessor Systems”, Term paper, EECS Department,
Milwaukee School of Engineering, North Broadway, Milwaukee, WI, USA, 1998-99.
[7] C. M. Krishna and G. S. Kang, Real-Time System, McGraw-Hill, 1997.
[8] K. Ramamritham and J. A. Stankovic, “Scheduling Algorithms and Operating Systems Support for Real-time Systems”, J.A
Proceedings of IEEE. Vol. 82, No. 1, pp. 55–67
[9] R. Mall, Real-Time Systems: Theory and Practice. Pearson Education, 2007.
[10] G. Bernat, A. Burns, A. Liamosi, “Weakly hard real-time systems”, IEEE Transactions on Computer Systems, Vol. 50. No.4, pp.
308-321, 2001.
[11] J. L. Diaz, D. F. Garcia and J. M. Lopez, “Minimum and Maximum Utilization Bounds for Multiprocessor Rate Monotonic
Scheduling”, IEEE Transactions on Parallel and Distributed Systems, Vol. 15, No.7, pp. 642–653, 2004.
[12] E. S. Kim, C. .S. Sung and I. S. Lee, “Scheduling of parallel machines to minimize total completion time subject to s-precedence
constraints. The Journal of Computers & Operations Research, Vol. 36, pp. 698–710, 2009.
[13] Y. k. Kwok and I. Ahmad, “Dynamic Critical Path Scheduling: an Effective Technique for Allocating Task Graphs to Multipro-
cessors”, IEEE Transactions on Parallel and Distributed Systems, Vol. 7, No. 5, pp. 506–521, 1996.
[14] R. .C. Correa, A. Ferreira and P. Rebreyend, “Schuduling Multiprocessor Tasks with Genetic Algorithms”, IEEE transactions on
Parallel and Distributed Systems, Vol. 10, No. 8, pp. 825–837, 1999.
[15] M. Yoo. “Real-time task scheduling by multi-objective genetic algorithm”, The Journal of Systems and Software, Vol. 82, pp.
619-628, 2009.
[16] M. Yoo, M. Gen, “Scheduling algorithm for real-time tasks using multi-objective hybrid genetic algorithm in heterogeneous
multiprocessors system”, Computers & Operations Research, Vol. 34, pp. 30843098, 2007.
[17] M. Yoo and T. Yokoyama, “Multi-objective GA for Real Time Task Scheduling”, Proceedings of the International Multi-
Conference of Engineers and Computer Scientists, Vol 1, IMECS 2016, March 16 - 18, 2016, Hong Kong, 2016.
[18] B. Malakootim, S. Sheikh, C. Al-Najjar and H. Kim, “Multi-objective energy aware multiprocessor scheduling using bat intelli-
gence”, J Intell Manuf, Vol. 24, pp. 805–819, DOI:10.1007/s10845-012-0629-6, 2013.
[19] J. H. Holland, Adaptation in Natural and Artificial Systems. Cambridge, MA, USA: MIT Press, 1992.
 Debanjan Konar et al. / Procedia Computer Science 131 (2018) 591–599 599
D.Konar et al. / Procedia Computer Science 00 (2018) 1–9 9

[20] A. J. Page and T. J. Naughton, “Dynamic task scheduling using genetic algorithms for heterogeneous distributed computing”,
Proceedings of the 19th IEEE/ACM International parallel and distributed processing symposium, Denver USA, pp. 1530–2075,
2005
[21] S. Wu. Annie and Han. u, “An Incremental Genetic Algorithm Approach to Multiprocessor Scheduling”, IEEE Transactions on
Parallel and Distributed Systems, Vol. 15, No. 9, pp. 824–834, 2004.
[22] M. R. Bonyadi and M. E. Moghaddam, “A Bipartite Genetic Algorithm for Multi-Processor Task Scheduling”, International
Journal of Parallel Programming, Vol. 37, No. 5 pp. 462-487, 2009.
[23] S. N. Sivanandam, P. Visalakshi and A. Bhuvaneswari, “Multiprocessor Scheduling using Hybrid Particle Swarm Optimization
with Dynamically Varying Inertia”, Int. J. Comput. Sci. Appl., Vol. 4, No. 3, pp. 95-106, 2007
[24] H. Chen and A. K. Cheng, “Applying Ant Colony Optimization to the Partitioned Scheduling Problem for Heterogeneous Multi-
processors”, Special Issue IEEE RTAS 2005 Work-in-Progress, Vol. 2, No. 2, pp. 11–14, 2005.
[25] M. F. Ercan, “A Hybrid Particle Swarm Optimization Approach for Scheduling Flow-Shops with Multiprocessor Tasks”, In:
Proceedings of the International Conference on Information Science and Security, pp. 13–16, 2008
[26] H. Mitra and P. Ramanathan, “A genetic approach for scheduling non-preemptive tasks with precedence and deadline constraints”,
Proceedings of the 26th Hawaii International Conference on System Sciences, pp. 556–564, 1993.
[27] M. Lin and L. Yang, “Hybrid genetic algorithms for scheduling partially ordered tasks in a multi-processor environment”, Pro-
ceedings of the 6th International Conference on Real-Time Computer Systems and Applications, pp. 382-387, 1999.
[28] Y. MonnierY, JP. Beauvais and AM. Deplanche, “ A genetic algorithm for scheduling tasks in a real-time distributed system”, In:
Proceedings of the 24th euromicro conference, pp. 708-714, 1998.
[29] J. Wu. C. Oh, “Genetic-algorithm-based real-time task scheduling with multiple goals”, Journal of Systems and Software, Vol.
73, No. 3, pp. 245258, 2004.
[30] M. R. Devi and A. Anju,“Multiprocessor Scheduling of Dependent Tasks to Minimize Make-span and Reliability Cost Using
NSGA-II”, International Journal in Foundations of Computer Science & Technology (IJFCST), Vol. 4, No. 2, 2014.
[31] M. R. Gary and D. S. Johnson, Computers and Intractability: A Guide to the Theory of NP Completeness. W.H. Freeman and
Company, 1979.
[32] M. A. Nielson and I. L. Chung, Quantum computation and quantum information, Cambridge University press, 2000.
[33] P. W. Shor, “Algorithms for quantum computation: Discrete logarithms and factoring, Proc. 35th Annu. Symp.Foundations of
Computer Science, Piscataway, NJ: IEEE Press, pp. 124–134, 1994.
[34] L. K. Grover, A fast quantum mechanical algorithm for database search, Proc. 28th ACM Symp. Theory of Computi, pp. 212–219,
1996
[35] A. Narayan and M. Moore, “Quantum-inspired genetic algorithms”, Proc.IEEE Evolutionary Computation, pp. 61–66, 1996.
[36] K. H. Han and J. H. Kim, “Genetic Quantum Algorithm and its Application to Combinatorial Optimization Problem”, Proc.
Congress on Evolutionary Computation, Vol. 1, pp. 1354–1360, 2000.
[37] K. H. Han, J. H. Kim. 2002. Quantum-Inspired Evolutionary Algorithm for a Class of Combinatorial Optimization. IEEE Trans.
Evol. Comput, Vol.6, No. 6, pp. 580–593, 2002.
[38] K. H. Han and J. H. Kim, “Quantum-inspired evolutionary algorithms with a new termination criterion, H gate, and two phase
scheme”, IEEE Transactions on Evolutionary Computation, Vol. 8, No. 2, pp. 156–169, 2004.
[39] K. H. Han and J. H. Kim, “On the analysis of the quantum-inspired evolutionary algorithm with a single individual”, In Proc. of
IEEE Congress on Evolutionary Computation. Vol. 2, Vancouver, Canada, pp. 91729179, 2006.
[40] K. H. Han and J. H. Kim, “Parallel quantum-inspired genetic algorithm for combinatorial optimization problem”, In Proc. of
IEEE Congress on Evolutionary Computation,Seoul, Korea,, Vol.2, pp. 14221429, 2001.
[41] B. B. Li and L. Wang, “A Hybrid Quantum-Inspired Genetic Algorithm for Multi-objective Flow Shop Scheduling”, IEEE
Transactions on Systems, Man and Cybernetics PART B: CYBERNETICS,Vol. 37, No. 3, pp. 576–591, 2007.
[42] D. Konar, K. Sharma, S. R. Pradhan and S. Sharma, “An Efficient Dynamic Scheduling Algorithm for Soft Real-Time Tasks in
Multiprocessor System Using Hybrid Quantum-Inspired Genetic Algorithm”, Proceedings of the 4th International Conference
on Frontiers in Intelligent Computing: Theory and Applications (FICTA) 2015, Advances in Intelligent Systems and Computing,
pp. 1-11, 2015.
[43] D. Konar, S. Bhattacharyya, K. Sharma, S. Sharma and S. R. Pradhana, “An improved Hybrid Quantum-Inspired Genetic Algo-
rithm (HQIGA)for scheduling of real-time task in multiprocessor system”, Applied Soft Computing, Vol.53, pp. 296307, 2017.
[44] D. Mcmohan,Quantum Computing Explained. John Wiley and Sons, Inc., Hoboken New Jersey, 2008.
[45] A. Arulselvan, C. W. Commander and P. M. Pardalos “A Random Keys Based Genetic Algorithm For The Target Visitation
Problem”, Advances in Cooperative Control and Optimization, 2007.
[46] A. Ruiz, M. A. Sambola, E. Fernndez, and M. G. C. Resende, “A biased random-key genetic algorithm for the capacitated
minimum spanning tree problem”, Computers and Operations Research, vol. 57, pp. 95–108, DOI: 10.1016/j.cor.2014.11.011,
2015.