Mosquito Detection For Improved Malaria Vector Modelling

The HumBug project is targeted at eliminating mosquito-borne diseases that have a major impact on human health, income and mortality [1]. The ultimate aim is to deploy a sensor network consisting of portable devices, such as mobile phones, wristbands or other monitoring devices. The devices may communicate in either a centralised or decentralised fashion to help target areas in need of aid optimally. The detection is critical to the success of the overall project in both configurations. This paper presents a simple, robust and highly computationally efficient wavelet-based algorithm. Classifier performance is demonstrated in demanding scenarios, showing equal or better detection performance at greatly reduced computational complexity compared to HumBug’s current detector. The improved efficiency particularly suits power needs of low cost devices and ensures sensor network policies can be utilised to their maximum potential.