Abstract
Image interleaving has proven to be an effective solution to provide the robustness of image communication systems when resource limitations make reliable protocols unsuitable (e.g., in wireless camera sensor networks); however, the search for optimal interleaving patterns is scarcely tackled in the literature. In 2008, Rombaut et al. presented an interesting approach introducing a packetization mask generator based in Simulated Annealing (SA), including a cost function, which allows assessing the suitability of a packetization pattern, avoiding extensive simulations. In this work, we present a complementary study about the non-trivial problem of generating optimal packetization patterns. We propose a genetic algorithm, as an alternative to the cited work, adopting the mentioned cost function, then comparing it to the SA approach and a torus automorphism interleaver. In addition, we engage the validation of the cost function and provide results attempting to conclude about its implication in the quality of reconstructed images. Several scenarios based on visual sensor networks applications were tested in a computer application. Results in terms of the selected cost function and image quality metric PSNR show that our algorithm presents similar results to the other approaches. Finally, we discuss the obtained results and comment about open research challenges.
Highlights
It is known that, when sending multimedia data through communication networks, the inherent presence of packet loss can lead to discernible quality degradations at the receiver side [1]
The problem of finding a packetization pattern with genetic algorithms was tackled. This technique allows one to engage the huge amount of possible solutions to the problem, when robustness against packet loss is required. This problem has been treated very little in the literature, but it has enormous application in image communication systems where resource savings are required
Patterns generated by the proposed Genetic Algorithm (GA) were evaluated using as the fitness function the cost function defined by Rombaut et al in [14], which evaluated a resulting packetization mask in accordance with three main properties
Summary
It is known that, when sending multimedia data through communication networks, the inherent presence of packet loss can lead to discernible quality degradations at the receiver side [1]. Sensors 2017, 17, 981 cases, even if unable to perform error correction, interleaving methods have proven to be effective on the reduction of the impact of network errors [5]. Due to its simplicity and effectiveness, various works have adopted image interleaving for enhancing image communication robustness. A similar approach is adopted in [8,17], where a chaotic torus automorphism is applied to enhance the robustness of a real-world wireless camera sensor network. As said, calculated coordinate ( x 0 , y0 ) is performed to select the block to be packetized. Even if such techniques are efficient and effective for enhancing robustness in a communication system, the search for optimal packetization patterns is still poorly studied. Most approaches behave always the same way, but there are no evaluation methods that give us a clue about, given a certain scenario, which interleaving pattern is the most suited
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
