SUBJECT CODE CS707
ASSIGNMENT NO 1
STUDENT NAME MUHAMMAD ADIL
STUDENT ID MS170401318
COURSE INSTDr NAHIL MEHMOOD
Modeling internet of things: a simulation perspective
Question No 1:
What problems discussed in research papers?
The problem discussed in this paper is about the internet of things. How this internet of thing can work. How it can deploy different kind of services on different kind of territories. This paper further more imposes the use of sophisticated modeling and simulation technique. Novel approach is discussed for the scalable simulation scenario, which enable large populated environment in real time execution. Highly demand simulation is achieved by the combination of novel hybrid, multi-level technique, agent base, adaptive parallel and distributed simulation.
We know that in the current era of world we are connected to internet by means of some devices like sensor, people equipped with mobile terminal. This means of connection can be consider the fast way of communication. It is therefore important to devise strategies to let them interconnect.
These considerations must be kept in account that things may have very specific characteristic both in term of hardware, software, and management utilities. Internet of things has equally importance for quantitative and qualitative aspect.
Question No 2:
How it is solved by different researchers
Answer No 2:
As per the conclusion of the research paper the existing network simulator are unable to support the required future of IoT system. That is why for improvement different research is suggested for solution.
For the improvement of simulator scalability, simIoT uses cloud environment for the implementation on back end operation.
For the urban environment the integration of general purpose discrete event simulation (DEUS) is setups with specific simulator (Cooja and ns-3). This was observed that the performance of integrated simulator have a good scalability even if the DEUS is java based and monolithic architecture.
Another integration simulator approach is proposed for hybrid simulation environment. In the hybrid (OMNeT++) framework the IoT network infrastructure is composed on sensor, actuator
Internet of Simulation (IoS) approach is suggest of interconnected simulation in all model.
SDL approach is considered the scalability for ns-3. The SDL is a language based IoT model used for automatic code generation.
Another very interesting solution is discussed is the integration between an agent-based methodology and domain specific (OM-NeT++). The paper highlighted that the agent based methodology is used for Modeling and OM-NeT++ for implementation of simulation.
Question No 3:
What are the issues with existing techniques?
There are many issues that should be consider in simulation point of view for Internet of things with existing techniques, one of them is most dominating that is called scalability. Scalability has effect on the amount of modeled entities and granularity of events. We knew that the IoT is consist of thousand of interconnected devices they have specific behavior and characteristic.
2- The other issue is to be consider in the Internet of thing prospective is the when we work with hybrid and multi level approaches for information exchange. This work on the synchronization among different model component in run time for information exchange. To overcome this problem the research paper suggests CloudSim for hybrid and multi level simulation architecture.
Question No :How it is solved by proposed work-and how it is different from others
The proposed worked is solved by multi- level Hybrid Simulation. This model has accuracy assessment of IoT have scalability problem with the large number of nodes. In multi-level Hybrid Simulation all the simulation is performed in the at Level 0 (with minimal detail). This means act as in-charge of all model components and their interaction. As discussed earlier in the paper 0 Simulator implement time stepped synchronization. When Simulation of a group is needed then the other group is triggered for simulation, this means that the simulation is of one group is transformed from level 0 to level 1. The result show that specific simulation is still working in Level 0 while some of them are transformed to Level 1. After studying this approach it is clear that different simulation tools can be used for different area. If needed this can be extended to Level 2. This procedure is managed under the time-stepped synchronization mechanism. This means all the interaction all interaction of level 0 component are performed at coarse Grained timestep. While the interaction on level 1 performing at fine Grained timestep. Finally the interaction between components managed at different levels that need to be arranged at the coarse grained timesteps.
In Multi-level Hybrid simulation the number of simulated entities is not changed, but the level of detail is used to perform the analysis. This increase the scalability of the simulation system. The detailed simulation is performed when it is necessary for the subset of entities. It should be clear that higher level simulation have lack of detail which may cause error. So the trade-off here become when and where in the simulated area triggering more detailed rather than simplified simulation, which has a great chance of error. That is why in every simulation appropriate verification and validation techniques is to be used.
How this Paper Different from others:
This paper is different from the other in the following scenario the important use case for internet of services relates to transportation system. The example for this include is BestMile: manage the autonomous vehicle fleet is a cloud platform. Kiunsys is deal with the developing solution regarding parking. It includes from analytics software to sensor management. In the city of Italy (La Spezia) installed more than 1000 parking spots they communicate in real time. Internet of Things architecture installed in future will be “cloud based” that will built a network of interactive modular sensor for collection of data in real time.
The methodology used in the given research paper is for scalable simulation scenario is the real-time execution of huge populated IoT environment including the important techniques that multi level hybrid simulation. Multi level hybrid simulation are importantly used with the integration of agent base, adaptive parallel and Distributed Simulation (PADS) approaches that can provide means to perform high in depth simulation on demand.
Detailed of simulation events and techniques are as follows:
Simulation and Discrete Event Simulation:
Parallel DES and PADS
The paper concludes and presents the issue regarding Internet of things while simulation and the deployment of smart services on smart territories. Further we studied about the two main issues like scalability and high level detailed in simulation.
We studied about different techniques and also to simplify the advantages of a hybrid, multi-level simulation approach. We also studied about the intelligent transportation systems. In intelligent transportation system method Wide Geographical Areas with a abundance of simulation entities that can be simulated with Agent-Based PADS.
Multi-Criteria Strategy for Job Scheduling and
Resource Load Balancing in Cloud Computing
Q1) What problems discussed in research papers?
The problem discussed in this research is about cloud computing. The requirement of Cloud computing is increase day by day and challenges faced by resources management. The research paper constant talking about the resources management, because of the different user demand, Quality of Service ( QoS), Load balancing and throughput. Cloud computing consists of Application, Infrastructure, storage, communication and other offered services. Could computing is an important element of Information Technology now a days. Cloud computing is heterogeneous and its resources are geographically distributed from problem of job to response time, availability and throughput.
Cloud computing can bring them all together for efficient utilization and according to the use requirement.
The other problems discussed in the research paper are about the addressing, load balancing in virtual machines, resources scheduling.
Question No:2 How it is solved by different researchers?
The following table represents different researchers they work on the solution of these problems:
Author/Year Method and Issue by Researcher Strength Weakness
Goyal T. and
Agrawal A. 2013 The use of Suffrage
Coupled with the Genetic Algorithm.
To minimize the make span of the jobs. The use of Suffrage
Coupled with the Genetic Algorithm Must be able to find minimum completion time for a given tasks. The use of Suffrage
Coupled with the Genetic Algorithm does not concern about
the Load Balancing between the virtual machines.
Gu J., Hu J. Zhao T.
and Sun G. 2012 Genetic algorithm in this research for the problems of Load balancing and high migration cost.
The algorithm in the research analyses the problems of load balancing in Virtual Machine and
resource scheduling. This technique is useful for the solution of
the problem load imbalance and high
migration cost. This research strategy does not take into consideration for the minimum completion time of a given tasks.
Katyal M. & MishraA. 2014 The given Algorithm using two conventional
task scheduling algorithm and this algorithm is based
on min-min and max-min conventional scheduling Techniques.
The given technique is used to reduce the overall makespan of the jobs. The algorithm given in this research ensures, that the jobs are executed with smaller delays to minimize the overall makespan. The algorithm made Not much improvement
Some improvement is still required to balance the load between the Virtual Machines and to
reduce the execution
Agarwal A. & Jain
S. 2014 The given algorithm i.e. Generalized Priority
Algorithm rank and the virtual machines
according to their Million Instructions Per Second (MIPS) and the cloudlets according to their sizes.
The aim of the research was to propose algorithm for scheduling that would help to schedule task efficiently in Cloud Computing Environment. The given Algorithm
was found better than in performance from FCFS and Round Robin The given Algorithm prioritized the job scheduling based on MIPS of VM and
cloudlets size without considering the completion time and load balancing.
What are the issues with existing techniques?
Following are the issues to be considered for the existing techniques:
If we observe the overall paper Multi-Criteria algorithm becomes complex for different user demand due to its distributed nature when the user demand arrive at the same time not much improvement has been made from FCFS. For the sake of balance the load between the VMs and minimize execution time still some improvement. The Algorithm further prioritized the job scheduling based on MIPS of VM and cloudlets size without considering the completion time and load balancing.
Question No:4 How it is solved by proposed work-and how it is different from others?
The problem is solved by the proposed work in the following manner:
The three scheduling techniques Min-min, Max – min and suffrage have great importance for the solution of the problem. They use the benefits from genetic algorithm for resource balancing. We seen that Genetic algorithm are helpful in job response time, availability of resource, throughput and the performance of cloud environment. The overall span-in space or time is reduced or minimized by the use of min-min and mix-mix algorithm.
“How this Paper Different from others”
The paper is different from others in sense that it avoids extreme utilization of the cloud resources by use of three scheduling techniques (min-min. mix-min and suffrage algorithm) and genetic algorithm. The main objective of this is to minimize the make span of the jobs and enhance job scheduling.
Methodology used in this research paper?
Methodology used for this paper consists on three stages for execution of jobs in the form of in RMK algorithm.
In stage 1: All the cloudlets information and virtual machines execution time information are contained in the stack table.
In stage 2: Three scheduling techniques (min-min, max-min and suffrage algorithm) are used to minimize the make span of the jobs.
In stage 3: In this stage we combine/encode the results from first two stages on Genetic Algorithm for resource load balancing using as a parameter.
At the end of that all the result shows, when the performance was tested on cloudSim tool kit. All techniques of RMK virtual machines with smallest MIPS are utilized more and have highest execution time. but once we observed it RMK case it was found reverse order and this order will increase the response time.
After studying this paper we are now able to conclude that this paper consists of algorithm for improvement of job scheduling and resources load balancing in cloud computing environment. One of the algorithms was RMK algorithm which has three scheduling techniques min-min, max-min and suffrage for its execution of this assigned task. Stack table contain all the information regarding the execution time of jobs in virtual machines and find minimum make span and load balancing using Genetic Algorithm (GA) and the three stages of this algorithm.
This increase the overall job scheduling and load balancing.