|
Description The Wi-Fi GRID is an implementation of existing over-the-counter routers from various brands such as Linksys, Belkin, Netgear and D-Link. The "GRID" part stands for Generic Redundant Integrated Deployment. This setup, which requires at least three routers, is designed to handle high client density in a small area in an automatic and falsely intelligent way. As many as 254 clients (as many as one DHCP/subnet will allow) are able to connect to one BSSID without dropping the connection. Increasing the number of identical routers improves throughput, redundancy, and overall reliability of the network. Such setups are suitable for local coffee shops, libraries, and completely wireless LAN. At least 3 Wi-Fi routers are needed. More routers increase the number of clients that can connect and ensure better load balance. At least 5Mbps down and 2Mbps up is recommended. Network Structure One router is connected to the main internet connection and functions as the sole DHCP server supplying IP addresses to other routers. Subsequent routers are connected to the DCHP router through their LAN ports with their individual DHCP functions turned off. Their Wi-Fi settings are configured to mirror that of the main router in terms of security (WPA/WPA2), and channel (1-11 in the US). Note that the WAN port is not used on the subsequent routers and they only act as a switch and access point. Why it works Each access point is mirroring the main and the client computers communicate via packets that contain Layer 2 data identifying only the DHCP server. The DHCP server returns packets at the same hardware level. Jumping between nodes does not interrupt where packets can be dynamically routed across all access points to the end-user client or server. Comparison to other network deployments Unlike a WDS (Wifi Distribution System) and mesh network, a GRID is only designed to operate in a small area and each node forms a direct connection to the client and does not make multiple hops between access point nodes to get to the DHCP server. Each server to client connection runs in parallel and the range of the network is only limited to the original range of each individual access point. The only improvement is reliability from the client hopping from node to node and joining less busy nodes as necessary. Nodes must be added manually and the system is not self-healing like a mesh network.
|
|
|