Hi,
I have to implement ABC algorithm in cloudsim for load balancing. I have initialized the population with no. of tasks and have considered half as employed bees and half as onlooker bees. By selecting a bee as scout bee the capacity of the virtual machines are calculated. But next I am not getting how to assign employed bees to virtual machines (is it randomly)? And also the fitness is calculated by
fitnessi=sum of task lengths of j/evaluate capacity of VMj(capacityj)
Here the single employed bee is assigned to a VM then what is the meaning of sum of task length of VMj? Can anyone please help me to clear my doubts?
Thanks in advance.
With Regards,
Kshama S B
Hello,
I don’t understand the question well but im guessing you’re trying to compute the fitness for a VM to host a submitted task. Task length of VMj is the sum of number of instructions of all the tasks running in VMj . Intuitively, the lower the load is on a VM, the higher the fitness value.
Best of luck
Thank you very much for the response. As I have told earlier once the population is initialized with the employed bees (50%) and onlooker bees (50%), the capacity of the VMs are calculated.
My question is whether the employed bees are assigned randomly to the VMs or based on any other criteria? And can more than one employed bee be assigned to the same VM?
With regards,
Kshama S B
I know few about the particular honey bee optimization approach.
In genetic and particle swarm after iterating through a fitness function or a velocity function, the best solution is found. It's up to you to modify fitness function and welcome a new fitness function based on your own idea or a well known optimization function like Rosenbrock
Regards,
The fitness function you mentioned here is to calculate the execution time of task which is generally:
fitness i = length of task i / capacity of VM j
Usually, if we want to calculate the execution time of workflow (which contains multiple tasks), we need to consider all tasks in this workflow. Otherwise, we consider single task.
Regards the initial population, you can initialise it randomly before the searching loop of ABC algorithm.
Based on your assumptions, you can decide whether a VM can host one task or more.
I Hope this will help you in your research.
Thank you Dabiah. Your answer really helps me in my research. Can you please help me to clarify some more doubts?
1. In the onlooker phase whether the onlooker bees are allocated only to the best fitness VMs? If so, if best fitness VMs are very few, are bees allocated to few VMs only? Remaining bees will they become scout?
2. Based on what criteria max cycle(loop) value can be decided? Can you please tell me the need of loop in the algorithm.
Your answer will help me a lot in my research. Thank you in advance.
With regards,
Kshama S B
Hi Kshama,
Sorry for the delay. I just saw your questions.
(1) Sorry, I do not know exactly how ABC works.
(2) The algorithm will be terminated if the best solution is found. In your case, the best VM that minimises the execution time of the task. Usually, after predefined number of maximum iterations, the algorithm will be terminated. For more accurate results, if the difference between the current solution in iteration t and the previous solution in iteration t-1 is
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Algorithms