The effective coverage distance of wireless LAN (WLAN) being small, users may leave the coverage area of the specific access point (AP) from time to time while roaming. However, the wireless network is a shared medium. The air is open for everyone. In general there is collision if a few users attempt to transmit with the same channel that is more rigorous during handoff period because of active scan mode. The active scan will perform requests for searching available AP. Unfortunately, this function consumes too much resource in wireless communication, and also affect total performance. We will propose an advanced active scan to improve it. In our proposal, we convert RF signal distribution to a simple classification problem, like as XOR classifier with artificial neural network (ANN). We combine ANN with active scan to achieve our goal. And the weight, which trained by ANN presents the connection character of geography. Moreover, the weight could be stored in AP for reusing and is called geographical fingerprint. The average enhancement of reducing the active scan area is about 62%.
Relation:
Wireless Communications and Mobile Computing 6(1), pp.127-135
