Pollination Algorithm Research Articles

Multi‐objective reliability‐based workflow scheduler: An elastic and persuasive task scheduler based upon modified‐flower pollination algorithm in cloud environment

SUMMARYThis research article formulates contemporary approach named multi‐objective reliability‐based workflow scheduler. Numerous strategies have been proposed in the past to prioritize and map the tasks to cloud resources. Though the recent studies lead to efficient solutions however they are restrained in terms of performance due to lack of resource consideration based on utilization rate and reliability index. It is crucial to consider reliability parameter while mapping tasks onto the virtual machines and not just the reliability value, but the cost incurred must also be minimized. To this end, the proposed strategy has been categorized into four modules, (i) scrutiny of reliable VMs, (ii) task ranking, (iii) optimizing the task re‐ordering using flower pollination optimization, and (iv) task mapping onto the VM. It intends to map task onto the most suitable machine in terms of makespan, efficiency, and incurred cost. In the experimental setup, four scientific workflows have been considered namely, LIGO, Genome, Cybershake, and Montage, they have been tested on the proposed approach while making comparison with the existing approaches namely FPA, GWO, and GA. The simulation results justified the claims by allocating resources to the cloudlets efficiently and stabilizing all the aforementioned parameters by attaining performance measures adequately.

Hybrid Algorithm Based Effective Light Trail Creation in an Optical Networks

Abstract With the overwhelming use of the Internet, the wavelength division multiplexing (WDM) optical networks have become the major mode of communication for wide applications. Hence, the routing and wavelength assignment (RWA) for a given traffic in WDM networks is the chief challenge in optical networks. This paper presents the RWA problem in WDM networks using the hybrid algorithm of flower pollination (FP) and simulated annealing (SA) algorithm (FA+SA) with and without wavelength conversion capability at the nodes of the network. Comparative analysis with techniques presented in literature is also carried out, i. e. differential evolution algorithm (DEA), firefly algorithm (FA), particle swarm optimization with less number of informers (Psolbest), SA and genetic algorithm (GA). The results prove that hybrid algorithm of FP and SA is potentially more efficient than algorithms present in literature for the light path establishment in an optical network.

Effect of 1070 nm laser intensity on parameters of In0.3Ga0.7As solar cell

The photoelectric properties of In0.3Ga0.7As solar cells applied in laser wireless power transmission (LWPT) were studied when they were irradiated by 1070 nm continuous wave (CW) laser of various intensities. The influences of laser intensity on solar cell parameters extracted by the pollination algorithm were analyzed quantitatively. Results show that the conversion efficiency of the cell rose to the maximum and then decreased rapidly in the laser intensity range of 50–900 mW/cm2. With higher energy laser irradiation, the rise of ideality factor and reverse saturation current would lead to the degradation of voltage at the maximum power point, which was the main reason for the decrease of conversion efficiency. The results provide the basis for choosing the appropriate input energy in the case of different transmission systems.

Online Social Networking services and Spam Detection Approaches in Opinion Mining- A review

Social network platforms have gained enormous growth in present days. Users are more attracted towards social network sites such as Twitter, Facebook, micro blogs and YouTube to express their opinion and views and share the knowledge. With the tremendous increase in technological growth and with the diversity of products, online e-commerce servicing sites are becoming competitor's for each other to increase their proliferation in business offering to post positive product spam reviews. Users generally express their opinion based on different sentiment orientations, ratings and the features of the product. This tends to create ambiguity at the customers end in arriving at a decision based on the criticism, building fake opinion on the products. Opinion spam detection provides scope in building opinion from the reviews of various networking sites. A novel framework using meta-heuristic and k-means clustering approach is proposed for identifying opinion spam detection using flower pollination, grey wolf and moth flame. Amazon automotive product dataset is chosen for analysis and it is observed that grey wolf algorithm performs better than flower pollination and moth flame optimisation algorithms in terms of improved convergence speed, mean, standard deviation, variance and elapsed run time.