dan_hebert
dan_hebert
dan_hebert
dan_hebert
dan_hebert

Into the mesh

Oct. 19, 2005
Senior Tech Editor Dan Hebert, PE, identifies the leading candidates for best system architecture supremacy in the wireless arena and believes software algorithms will be key for high bandwidth mesh networks.
By Dan Hebert, PE, Senior Technical Editor

THE WIRED fieldbus wars continue apace, but at least it’s possible to identify the leading candidates for supremacy. Not so in the wireless world, where rapid technological advances hinder standardization efforts. One of the first questions to be answered in the wireless arena is, “What is the best system architecture?” Point-to-multipoint and mesh are the leading candidates. The betting here is that mesh will emerge victorious because only mesh allows alternate communication paths to be configured automatically as needed. This is critical with wireless because paths are often blocked because of interference, especially when some of the mesh nodes are mobile.

If mesh is the winner, the next question is, “What type of mesh?” As in the wired world, there are two main types of data communication required for wireless process control and monitoring applications. The first is device-level communication characterized by many small data packets. The second is higher-level communication characterized by comparatively fewer data packets with each packet containing more information.

Device-level networks may be best served by the ZigBee standard. ZigBee Alliance members, including process control heavyweights Honeywell and Invensys, are defining a global specification for wireless applications based on the IEEE 802.15.4 standard. ZigBee takes full advantage of IEEE 802.15.4 and adds logical network, security and application software.

Higher-level networks with more data communications need more bandwidth, and this is the area where technical advances are rapid and standards are few.

Mesh networks typically use off-the-shelf radios and communicate via standard protocols. “Like most mesh network providers, we use standards like WiMax, licensed radio frequencies, and 802.11b, g and a,” says Richard Lander, the director of LocustWorld Ltd..

High-bandwidth mesh network customization takes place at the software level, and these software algorithms can be fiendishly complex. Challenges include routing, security and user connectivity.

Routing is particularly difficult in a mesh network because there can be many nodes and there are no pre-defined communication paths. Routing software must constantly monitor all network paths and automatically choose the correct path for each data packet. These paths change constantly due to environmental, network loading and node mobility.

Mobile nodes are especially difficult to incorporate into routing schemes. “For mobile node applications, we’re working on software to constantly monitor the mesh, anticipate best paths, and allow seamless path switching,” claims Bob Osann, CEO and Sales VP of Mesh Dynamics.

Another routing challenge is to recognize and automatically configure new nodes as they’re installed. This automatic configuration must be accompanied by stringent authentication, so that only registered devices are allowed on the network.

The basic architecture of mesh networks places unique demands on node hardware and creates a huge amount of network traffic because each node must be able to perform three types of communications: communication with a client, receiving network data, and sending network data.

Node clients can be data sources or data consumers. Data sources may be sensors or data concentrators. PCs and mobile phones can be both data sources and data consumers. If nodes are few and data is limited, these communication requirements are easy to manage, but adding many nodes while simultaneously increasing data throughput poses problems.

Because each node typically has one radio with one communication channel, nodes cannot send and receive data at the same time. This means that bandwidth is reduced with each hop, since each data packet has to be relayed. This results in a bandwidth loss of up to 50% per hop, depending on the mesh topology.

Four hops away the bandwidth would be 1/2 x 1/2 x 1/2 x 1/2 = 1/16 of what is available at the host. Data communications with node clients further degrade bandwidth. This bandwidth degradation means that mesh networks have difficulty scaling when data communication rates are high and nodes are many.

One way to deliver added capacity is to add more radios and more radio channels. Mesh Dynamics calls its high-bandwidth mesh network a Three-Radio Structured Mesh. One radio communicates with the client, another receives data, and a third sends data. Three separate radio channels are available for client communications, and up to eight separate channels are available for data send and receive operations.

The data send and receive radios operate independently and channels are allocated automatically to minimize interference. The client-communication radios are also in a different band (802.11b/g vs. 802.11a) than the data-send-and-receive radios to reduce contention. Bandwidth degradation effects endemic to single radios are eliminated because each client radio operates independently and simultaneous send/receives are allowed.