International Conference on Computing, Communication and Automation (ICCCA2015)
Review of Nature Inspired Algorithms in Cloud Computing Ritu Kapur Department of Computer Science NITTTR, Chandigarh, India
[email protected] Abstract—Cloud Computing is a major area of research. Nature Inspired Algorithms (NIAs) form the major portion of research going on in the Cloud today. NIAs as the name suggests are the algorithms whose source of inspiration is nature. NIAs can further be classified into algorithms based on Swarm Intelligence (SI), Biological Phenomena (called Bio-inspired BI), Physics and Chemistry systems or based on some other things. SI based algorithms are called intelligent because they are known to learn and improve their performance by observing the output on previous moves made by them. NIAs provide an efficient solution to many real-world optimization problems which are categorized to be NPHard Problems. NIAs have a huge list of applications and most of them prove to be more efficient than other algorithms and thus are many a time used in combination to other algorithms in order to improve performance and thus deliver a better QoS. The paper intends to review, classify and briefly describe various NIAs and the principle behind each algorithm so as to inspire further research. The paper also lists the applications of various NIAs. NIAs also find their application in Cluster and Grid Computing.
Keywords— Cloud Computing; QoS; Nature-Inspired Algorithms; Swarm-Intelligence based Algorithms; Bio-Inspired Algorithms; Optimization Problems; Meta-Heuristic-Algorithms; Cluster computing; Grid-Computing I.
INTRODUCTION
A substantial amount of genius solutions and algorithms are gestated in the nature, and we just need to dig them out, and then employ them to solve our problems [1]. The algorithms bear their source of inspiration from nature is known as Nature-Inspired Algorithms (NIAs). A. Nature Inspired Algorithms: NIAs are the algorithms inspired from any natural phenomena (biological or non-biological) or the behaviour of natural algorithms. Examples of NIAs are Ant System (AS), Ant Colony Optimization (ACO), Simulated Annealing (SA), Black Hole Algorithm, Water Cycle Algorithm etc. NIAs are subcategorized as follows [2]: •
•
Bio-Inspired Algorithms (BI) o
Swarm-Intelligence based Algorithms
o
Not based on SI
Physical and Chemical Systems
ISBN:978-1-4799-8890-7/15/$31.00 ©2015 IEEE
•
Other Algorithms
The categorization of these algorithms is given in figure 1 below represents their representation in set notation. As shown by figure 1, NIAs form the superset of all the listed algorithms. NIAs can thus be broadly classified as follows: 1) Bio-Inspired (BI) Algorithms: The largest fraction of NIAs is formed by the Biologically-Inspired (BI) Algorithms or Bio-Inspired Algorithms. The Bio-Inspired Algorithms are the algorithms which are inspired from any biological phenomena or from natural organisms. 2) Swarm- Intelligence (SI) based Algorithms: The BioInspired Algorithms which are inspired from the Swarm behaviour or Swarm Intelligence are classified under the SI category. “Swarm” is a general term used to refer to any restrained collection of Interacting Agents or individuals [3]. Examples of SI-based algorithms are Ant System (AS), Ant Colony Optimization (ACO), Bee system (BS), Bee Colony Optimization (BCO), Artificial Bee Colony (ABC) Optimization, Particle Swarm Optimization (PSO), Fireflies Algorithm (FA) etc. 3) Non SI based BI Algorithms: All the Bio-Inspired (BI) Algorithms do not come under the category of SwarmIntelligence (SI) based algorithms. Examples of such algorithms are Genetic Algorithms (GA), Simulated Annealing (SA) Algorithms, and Differential Evolution (DE) Algorithms etc. Although it is not always easy to classify all algorithms as SA and DE are not directly based on BI algorithms but since their working is similar to GA they are categorized into the category of BI algorithms. 4) Physics and Chemistry Based Algorithms: The Algorithms which are NIAs but not BI are classified under the category of algorithms inspired from Physics and Chemical systems. Examples of such algorithms are Water Cycle Algorithm, Black Hole, Spiral Optimization, and Stochastic Diffusion Search etc. 5) Other Algorithms: There are also some other algorithms which are nature inspired but do not come under the category of BI based or those based on Physics and Chemistry based systems. Examples of such algorithms
589
International Conference on Computing, Communication and Automation (ICCCA2015) applying the state transition rule and the initial pheromone deposition takes place. During their tour, the ants also update the pheromone on visited edges using the local pheromone updating rule. i. State Transition Rule: This rule determines the next node to move to. The state selected is the decision made on the bases of both the accumulated prior knowledge and the exploration of new edges.
are those based on music or different sources like emotional and social etc. Other Algorithms
NIAs
Bio-Inspired Physical and Chemical Systems
SI
s= {arg max(ij) Fig. 1.
{ S,
Set Classification of Nature Inspired Algorithms (NIAs)
In accordance to the figure 1 above, we can state the following [2]: SI
BI
NIA
(1)
RELATED WORK
τij(k)= ρτij(k-1) + (1-ρ)τ0
A. Ant Colony Optimization Algorithms (ACO) M. Dorigo [4] presented the basic ACO algorithm, Ant System (AS), in 1992. In 1997, M. Dorigo et al. [5] proposed Ant Colony System (ACS) as an improvement of AS [4]. According to the authors in [4- 6] ants have a general property of locating their path to food. In this phenomenon of locating their food, ants keep on depositing a chemical substance known as pheromone, which also keeps on evaporating with time. This pheromone in turn helps new ants on their path to find the shortest route to the food. The path having largest amount of pheromone represents the shortest route. So this process thus leads the new ants to their location in minimum time through the shortest path.
c)
The desirability of the edge (i, j) is inversely proportional to its length
Q
otherwise
(2)
where q0 is the probability of the best state being chosen and thus 0
