THE RIPHAH INTERNATIONAL UNIVERSITY, LAHORE

INSTITUTE OF COMPUTING

Project Management
Name Shahbaz Akhtar Javed
Program MSCS 4th
SAP ID 30330
Assignment 03
Submitted To Dr. Muhammad Yaseen
Research Article 01: Joint Optimization of Task Scheduling and Image Placement
in Fog Computing Supported Software-Defined Embedded System

1. Problem statement

To this Paper, three issues are investigated:

 How to balance the workload on a client device and computation servers, i.e., task
scheduling,
 How to place task images on storage servers, i.e., resource management, and
 How to balance the I/O interrupt requests among the storage servers.
2. Objectives

In this paper, we consider a software-embedded system supported by fog computing,

where images are stored on a storage server and calculations can be performed on an
embedded device or computer server. . Designing an effective task scheduling and resource
management strategy with minimum task completion time is critical to improving user
experience. To deal with the high computational complexity, an efficient computational
solution based on our formulation is proposed and based on extensive simulation.

3. Research method/applied model or technique

Because of its high complexity, MINLP problems are difficult to solve directly. To solve this
problem, we propose a three-stage algorithm based on the MINLP formula. The main
concept of this algorithm is division and integration.

 Minimize the I/O Time

 Minimize the Computational Time
 Joint Optimization
4. Size of requirements considered

An effective method of resource management and task scheduling is required to provide a

superior user experience such as reducing task completion time. Although remote I/O
interrupts are required, we can still achieve high performance by enabling fast compute
services on the client.
Research Article 02: Architectural Support for Task Dependence Management
with Flexible Software Scheduling

1. Problem statement

This article presents a concurrent hardware/software mechanism (Task Dependency

Manager) (TDM), which accelerates time-consuming tasks on a system running on
specialized hardware and allows flexible task scheduling policies in software. TDM extends
the minimal ISA so that the operating system can handle job creation, job dependencies, job
interrupts, and job efficiency. At the architectural level, TDM introduces a Dependency
Management Unit (DMU) that maintains a set of tables and lists about running tasks and
the dependencies between them. Ready-to-run tasks are exposed to the independent
runtime system to enforce software scheduling policies.

2. Objectives

The main contributions of this paper are:

 Newly designed hardware/software engine for rapid deployment and dependency

tracking while supporting flexible software applications. Hardware development
incorporates new architectural techniques with advanced proposals to reduce diversity
in architectures and reduce hardware costs.
 Detailed evaluation of TDM in a complete system simulation that includes application,
runtime system, operating system, and architecture layers. On a 32-core processor,
TDM showed a 12.3% average speedup and a 20.4% reduction in energy delay product
(EDP) compared to the software-based baseline.
 Demonstrate the capabilities of TDM when integrated with five software programs that
utilize different manufacturing processes. Thanks to this flexibility, TDM increases
overall hardware-based runtime by an average of 4.2%, EDP improves by an average of
6.2%, and footprint decreases by 7.3x.
3. Research method/applied model or technique
Task-based scheduling models are very attractive for large scale multicore. An important
aspect of job-parallel scheduling is job granularity, which determines the ability to use
available parallelism and ensure load balancing, but also determines the highest cost of
system execution time. This paper proposes TDM, a hardware/software based mechanism
to accelerate application dependency management processes while allowing flexible
application scheduling in software.

Unlike in the past that was fixed by programmers and architecture, the separation of
concerns and TDM provides a high degree of flexibility, flexibility and hybrid power, which
can be seen in modern equipment and many sockets and off- and shelf accelerators. cracks
etc. can benefit from it. Among the benefits of other policies is to schedule work and
activities.

In addition, the TDM architecture supports new features to increase efficiency, such as
redeploying identifiers to reduce required storage space or setting cache size to avoid
infrastructure interference during backups. The program address must be the same. As a
result, TDM increases process speed by more than 12.3% and reduces EDP by 20.4%.
Compared to pure hardware improvements, TDM achieves an average speed of 4.2% and
hardware complexity of 7.3x.

4. Size of requirements considered

All of these features require an integrated network to perform complex operations on the
deep data systems of the operating system. In addition, the changes provided by the
programming software are very important in the high-performance computing system and
the fast processing in the multi-space processing, which can improve the performance and
energy efficiency, and restricts the scheduling of applications and data. Animals require
software installation. Second, the memory requirements of the index can be reduced
because the size of internal identifiers is smaller than the 64-bit identifiers used in real-time
systems.
Research Article 03: A Bayesian Critical Path Method for Managing Common
Risks in Software Project Scheduling

1. Problem statement

The purpose of this paper is to find the best model for uncertainty management / risk
management in software design considering the scheduling and risk aspects of software
design.

2. Objectives

Research improved CPM-based scheduling by using BN to help software teams analyze

schedules and predict their probability of success, which effectively accounts for
uncertainty.

This method makes it possible to capture different areas of risk and use them to analyze the
software development process. It also shows the duration of each step and the uncertainty
of the overall process and probability distribution.

This approach provides the best way to deal with uncertainties that cannot be controlled by
traditional methods, such as uncertainty about the relationship between events and risks. It
also improves the quality of the software development program by ensuring that all key
aspects of the design (for example, dependencies, capabilities) are provided at a low level in
a mathematical model.

3. Research method/applied model or technique

 The RBCPM Method

The tool has the following main functions:

o Model and calculate a project schedule

o Alert project managers about tasks could be over-scheduled
 o Calculate the possibility of the schedule
o Calculate the possibility of each task.
o Model the network of risks that have impacts on each task.
4. Size of requirements considered

Need RBCMP Tool to perform this research.

Research Article 04: An Evolutionary Hyper-heuristic for the Software Project
Scheduling Problem

1. Problem statement

Our innovations are as follows: (1) this is the first work on evolutionary hyper elasticity
optimization for SPSP. (2) This is the first step for adaptive selection of crossover and
mutation operators. (3) We develop different validation methods for two types of
operators: translation and crossover. And (4) we use a multi-armed bandit strategy to select
crossover and mutation operators. We use a 3-size crossover pool and a 3-size transfer
pool. Our experiments show that our approach is effective in selecting crossover and
mutation operators for SPSP.

2. Objectives

This paper proposes an evolutionary Hyper-heuristic for solving SPSP. Hyper heuristic EA
uses a high-level strategy that automatically adapts mutation and crossover operators
during its evolution. A sliding window MAB strategy is used to adaptively search both
operators. Experiments on a set of 48 reference examples show that the proposed
algorithm can improve the SPSP and select strategies using simple random selection of
operators and advanced methods from the literature.

3. Research method/applied model or technique

We conducted a series of experiments to determine the effectiveness of the proposed AOS

technique for obtaining a simple random selection of crossover and mutation operators. It
also allows us to compare the results of this algorithm with those of the mode. We refer to
the three algorithms as GA-SMAB, GA-randAOS and GA for convenience. We conducted our
experiments on a reference data set used in 48 cases.

For each example, the average results of 30 independent runs of the algorithm are shown.
To minimize the variety of parameters, we choose the same parameters as used in the
literature, except for the translation probability, which is set to a value that guarantees the
performance of the translation operator.

4. Size of requirements considered

Need Hyper-heuristic Framework to perform this research.

Research Article 05: Scheduling Real-Time Parallel Applications in SaaS Clouds in
the Presence of Transient Software Failures

1. Problem statement

Here, we investigate the impact of fine-grained parallel application scheduling strategies in

a SaaS cloud on transient software issues that can occur during application implementation.
We integrate and compare different simulation techniques that include checkpoints and
program-defined approximations to: (a) provide resilience to transient software failures, (b)
ensure that all programs meet deadlines, (c) to deliver program outcomes that are high
quality and ( d) reduce financial costs for end users of approved applications. To the best of
our knowledge, none of the scheduling methods presented in the literature addresses all of
the above challenges in SaaS cloud environments.

2. Objectives

In this paper, we explore different strategies for organizing fine-grained parallel applications
in the SaaS cloud when software issues arise. Through simulations under different workload
conditions, we integrate and compare different techniques in program-defined areas and
approximation calculations, with multiple objectives of fault tolerance, quality of service
and reduction of end-user bills.

3. Research method/applied model or technique

 EDF
 EDF RAC
 EDF FAC
 EDF ADC
 EDF ADC RAC
 EDF ADC FAC
4. Size of requirements considered

SaaS cloud performance was assessed in a series of simulation runs in the implementation
process of each research group. To this end, our discrete event simulation program in C++ is
tailored to the needs of the specific problem given the complexity of the system and the
workload model under consideration. We use synthetic methods to reduce the uncertainty
of the simulation.