Design And Implementation Of Human Crowd Density Estimation System With Energy Harvesting In Wireless Sensor Network Platform

• Main Author: Fadhlullah, Solahuddin Yusuf
• Format: Thesis
• Language: English
• Published: 2017
• Subjects: T Technology; TA401-492 Materials of engineering and construction. Mechanics of materials
• Online Access: http://eprints.usm.my/45811/1/Design%20And%20Implementation%20Of%20Human%20Crowd%20Density%20Estimation%20System%20With%20Energy%20Harvesting%20In%20Wireless%20Sensor%20Network%20Platform.pdf; http://eprints.usm.my/45811/
• Institution: Universiti Sains Malaysia

Kepadatan yang tinggi dalam khalayak ramai boleh menjadi berbahaya kerana wujudnya potensi untuk pergerakan sekumpulan manusia secara tiba-tiba yang menyebabkan rempuhan dalam kes kecemasan. Untuk mengurangkan kecederaan mahupun kehilangan nyawa dalam kemalangan yang berkaitan dengan isu kepadatan manusia, sistem pengawasan kepadatan manusia berdasarkan frekuensi radio telah dibangunkan sebagai satu alat keselamatan. Sistem yang didapati pada masa kini mempunyai keupayaan pengawasan yang terhad; saiz pengawasan khalayak yang rendah, jarak pengesanan yang rendah, keperluan bilangan alat komunikasi yang tinggi dan jangka hayat operasi yang terhad. Faktor-faktor ini memberi kesan secara langsung kepada unsur praktikal dan ketepatan sistem penganggaran kepadatan manusia tersebut. Untuk mengurangkan kelemahan keupayaan pengawasan, satu sistem untuk mengesan kepadatan khalayak diusulkan berdasarkan kepada teknologi ZigBee dan rangkaian pengesan tanpa wayar yang meningkatkan jarak pengesanan khalayak kepada 30 m dengan hanya satu nod diperlukan setiap 37.5 m2. Hal ini dicapai tanpa mengurangkan bilangan khalayak (50 orang) yang boleh dikesan oleh sistem. Untuk menambahbaik ketepatan anggaran, kesan khalayak terhadap isyarat diselidik menggunakan kaedah statistik ‘One-way Analysis of Variance’ dan ‘Design of Experiments’. Hasil dapatan mengesahkan saiz khalayak memberi kesan yang paling besar terhadap kelemahan isyarat. Untuk interaksi di antara sifat-sifat khalayak, didapati saiz khalayak bersama bilangan alat penerima dan bentuk khalayak bersama bilangan alat penerima memberi kesan signifikan terhadap kekuatan isyarat. Faktor-faktor ini kemudian dimasukkan ke dalam algoritma H-CDE yang diusulkan. Algoritma pengesanan khayalak ini dan pengelasannya menunjukkan purata sebanyak 71.2 peratus ketepatan dalam mengenalpasti tahap kepadatan khalayak yang juga dapatan terbaik berbanding algoritma lain. Untuk mengatasi masalah kuasa yang terhad, mekanisma tuaian tenaga solar diperkenalkan ke dalam sistem H-CDE untuk memanjangkan jangka hayat operasi pengawasan. Kajian menunjukkan mekanisma tuaian tenaga ini mampu untuk memanjangkan operasi sistem pengawasan secara berterusan jika sistem ini mendapat paling kurang 5 hingga 6 jam pendedahan kepada sinaran matahari setiap 33 jam kitaran. Sumbangan kajian ini ialah pada penambahbaikan sistem berdasarkan teknologi frekuensi radio untuk mengesan kepadatan khalayak, penambahbaikan pada ketepatan penganggaran kepadatan khalayak yang didokongi oleh analisis statistik dan lanjutan operasi sistem melalui mekanisma tuaian tenaga. __________________________________________________________________________________ A crowd with high density can be dangerous due to the potential of a sudden surge of large moving bodies causing stampede in cases of emergencies. To mitigate casualties in crowd-related disaster, radio frequency-based crowd density estimation and monitoring system is being developed as a safety tool. Current systems have limited monitoring capabilities; low size of crowd monitored, low detection range, high number of transceivers required and finite operational lifetime. These factors directly influence the practicality and prediction accuracy of the system. To mitigate the limited sensing capability, a human crowd density estimation (H-CDE) system based on ZigBee and wireless sensor network technology is proposed that increases the crowd detection range to 30 m with only one transmission node required every 37.5 m2. This is achieved without sacrificing the amount of crowd detectable by the system (50 people). To improve the estimation accuracy, the effect of crowd on signal propagation is investigated using One-way Analysis of Variance and Design of Experiments statistical methods. The results verified that the crowd size significantly affects the signal attenuation. In the interactions between the crowd properties, crowd size * number of receiver and crowd pattern * number of receiver were found to significantly affect signal propagation. These factors are then integrated into the proposed H-CDE algorithm. The H-CDE algorithm and its crowd classification yielded an average of 71.2 % accuracy in identifying the level of crowd density, which is the best compared to other algorithms found in the literature. To solve the finite power problem, a solar energy harvesting mechanism is introduced into the H-CDE system to extend the operation of the monitoring system. It is demonstrated that the proposed energy harvesting mechanism could operate perpetually, given that the system is exposed to good sunlight at least for 5 to 6 hours in every 33-hour cycle. The contribution of the research is on the improved RF-based crowd density detection system, improved crowd estimation accuracy which is backed by statistical analysis and extension of its operations through the energy harvesting mechanism.