Research On Energy Efficient Image Transmission Algorithms In WMSN

Thanks to the great demand of application,much attention has been paid to Wireless Multimedia Sensor Network(WMSN)in recent years.However,since the sensor nodes are constrained by the resources such as limited energy,as a result,the energy efficiency needs to be improved in mass data transmission,e.g.images and videos.Mainly focusing on Compressive Sensing(CS),SoftCast,Cloud-Matching and joint optimization,this paper investigates the high energy efficient image transmission method in WMSN.The main works are as following:To solve the wireless image transmission problem of low energy consumption in an end-to-end model,a method based on adaptive block compressive sensing and SoftCast is proposed in this paper.Firstly,according to the receiving image quality,the relationship between sparsity and compressive sensing sampling rate of each image block is analyzed.Secondly,based on the SoftCast model,we analyze the relationship between the scale factors of compressive sensing measurements and energy consumption with some expected distortion,and the mathematic model of "sampling rate-scale factors-energy" is built.Lastly,the joint method is used to optimize the sampling rate and scale factors of each image block simultaneous for energy saving in wireless image transmission.The mathematic experiment results show that,comparing with the reference scheme,the energy consumption is lower with the approximately receiving image quality.To solve the problem of wireless image transmission from mobile terminal to the cloud,based on the fact that there has already existed lots of correlated images in the cloud,we improve the existed digital-analog transmission scheme,and proposed an energy efficient image transmission scheme based on digital-analog transmission and Cloud-Matching.Firstly,the distortion of side image is analyzed which caused by error bits when transmitting,meanwhile,the energy consumption of the system is analyzed when different transmission power factors are allocated for different frequency components of side image,and,the distortion-energy model is built for side image transmission.Secondly,according to the gradient descent algorithm,the power factors are optimized for energy saving with the expected receiving side image quality.Thirdly,when receiving the side image,the scale invariant feature transformation is extracted for correlative images searching in the cloud,which used for side image inpainting.Simulation results show that comparing with the reference scheme,the image quality is better with the proposed method at the approximate level of energy consumption.Specially,in the high noise environment,the receiving image quality can be improved much more.