Wireless Mesh Expands The Boundaries Of Networking

Sept. 15, 2005
Greater range and reliability, plus much lower costs, have businesses scrambling for wireless mesh to implement into their short-range wireless setups.

Most computers are on a network nowadays. Consumer electrical, electronics, and communications products are next. Industrial sensors and controls, machine tools, and other devices in process control and manufacturing are under way. And don't forget about machine-to-machine (M2M), which will let any machine talk to any other machine via several layers of networking.

One day, everything will be networked. That day may be closer than we think thanks to wireless mesh networking. With it, designers can interconnect any device with an inexpensive short-range wireless chip to everything else. This really opens the door to all sorts of new applications that simply weren't possible before.

MESH CONCEPTS In a mesh network, the nodes are all connected to one another. This is called a full mesh (Fig. 1). Each node has a direct link to all other nodes, making for a very useful arrangement. But as the number of nodes increases, the number of links becomes impractical. The number of links (L) is defined by the number of nodes (N) with the expression:

L = N(N - 1)/2

Connecting 20 PCs would require 190 links— a hardware and wiring nightmare. So, practical networks use a bus, ring, star, or other network topology plus some access method to reduce the number of interconnections. The only widely utilized wired mesh network is the Internet.

Going wireless makes the mesh usable and affordable. Furthermore, a partial mesh like the mesh in Figure 2 achieves the primary benefits of the full mesh. Here, not every node is connected to every other node. But if there are enough links, designers can realize some amazing advantages.

Mesh networks are primarily used for monitoring and control operations. As a result, the mesh usually carries sensor readings or control signals. Voice and video typically aren't involved, though some advanced mesh networks make that possible. Because monitor and control operations involve relatively short and simple packets, data rates can be very low. A few kilobits per second is often fast enough, and higher speeds can be implemented if needed. Typical data rates range from 20 to 250 kbits/s.

Perhaps the key benefit of the partial mesh is that the range of each node is greatly multiplied. Most short-range wireless technologies have a typical maximum range of 10 m or less. But this needn't be the maximum communications distance, since all other nodes are used as repeaters or routers.

A signal can be passed from node to node, extending the range indefinitely. In Figure 2, node A can get a message to node L by passing the signal through nodes A-B-E-M-I-L. An alternate path is A-C-D-F-L. There are several other multiple redundant paths, too. The data to be transmitted is put into a packet, and the packet is "hopped" from node to node until it reaches its destination.

In most applications, the nodes attempt to get data to a collection point or access point such as M in Figure 2. The data then is aggregated and sent on to a local-area network (LAN), metro-area network (MAN), or wide-area network (WAN) for further transmission (such as the LAN in a manufacturing plant or the Internet).

A popular variation of the mesh topology is a hybrid consisting of several point-to-multipoint (PMP) star networks, where multiple nodes talk directly to a central coordinating node or access point (AP). Then, the multiple APs are connected in a mesh configuration.

The fact that there's more than one path through a mesh introduces the other major benefit—reliability. If one path fails because of obstructions in the signal path, a defective node, or multipath attenuation, the signal can find one or more alternative routes. If the battery in a node fails, it drops out of the network, but other nodes relay the data via alternative hops.

Movement of people, vehicles, or equipment occasionally may block a previously good wireless path. Furthermore, temporary interference from another source or a sudden noise burst may prevent transmission. Again, a mesh automatically finds another path.

The total number of nodes is an important consideration in a mesh network. To benefit from multiple hop paths, many nodes are required. The absolute minimum mesh configuration is three nodes. Yet adding more nodes will significantly increase the reliability and robustness of the mesh.

Mesh networks also scale well. Initially, they may consist of only a dozen or so nodes. But that can scale to hundreds or even thousands of nodes without difficulties.

In addition, mesh networks are self-forming. The nodes automatically discover one another and establish a link if they're within range. This is called an ad hoc network. If the nodes are mobile, the network constantly and automatically reforms itself to the participating nodes.

A new node can be added at any time. If that node is too far away to link with existing nodes, additional repeater nodes may be added in between to establish a link. With small, inexpensive nodes, this approach is still cheaper than wiring in most applications.

POWER SAVINGS Mesh uses very little power, too. Because the distance between nodes is kept short, the transmit power needed to establish reliable communications is very low. In fact, some nodes may be battery operated.

Since nodes transmit packets in bursts, the node may go to sleep and draw only microamps, waking up only when called upon to relay a message or when it has one to send. Duty cycle may be only 0.1% to 1.0 %, greatly reducing power consumption. Battery life can last from many months to many years, reducing the need for frequent maintenance.

While self-organizing, self-healing mesh networks offer massive benefits, they also have a downside, namely security. Mesh networks can be hacked and compromised if they aren't protected. Yet protection is available with encryption methods like the Advance Encryption System (AES).

Another disadvantage for some applications is latency. Nodes take a finite amount of time to wake up and transmit data. Also, each hop requires a finite time. Total latency between nodes can be 5 to 30 ms. In some deterministic industrial control applications, this may not be fast enough. But in many instances, that latency isn't an issue.

THE RADIO INTERFACE Given so many available single-chip wireless transceivers today, which one is best for a mesh? The answer lies with the application. For instance, there's no reason why designers can't use the inexpensive ISM-band (industrial, scientific, medical) ICs operating at 315, 433, and 915 MHz. Bluetooth is another possibility.

When more inexpensive Ultra-Wideband (UWB) transceivers become available next year, the door will open for very highspeed, short-range mesh networking. Consumer electronics can use it to connect all of the various video and audio components together around the house. Maximum data rates for UWB, in the form of wireless USB or direct-sequence UWB, are 480 Mbits/s and 1 Gbit/s, respectively.

A mesh and repeaters can maintain the data rate over a longer range. Artimi and other companies are beginning to provide the software in an external embedded controller to create a mesh solution. One good possibility for a radio interface for mesh is the ubiquitous 802.11 Wi-Fi transceiver. The cost is very low and the data rate is high, from 11 to 54 Mbits/s. On the other hand, power consumption is high, and this standard implements a point-topoint (P2P) or point-to-multipoint (PMP) star topology.

Yet with the proper software and sufficient power, Wi-Fi is a good option if high data rates are needed. Several companies make software that converts a Wi-Fi radio into a mesh node. The IEEE is working on the 802.11s mesh-networking standard. A full standard isn't expected for several more years, though. In the meantime, several proprietary mesh systems are available.

Perhaps the best option is the newer wireless standard, IEEE 802.15.4. Also known as ZigBee, this standard was created from scratch to work in a mesh configuration. It defines the physical and data link (MAC) layers as well as the basic topology and interoperability between nodes (Fig. 3). The ZigBee Alliance has created the network and security upper layers, and it may develop application profile layers.

IMPLEMENTING MESH NETWORKS Many 802.15.4 chips are available if you're the do-it-yourself type. ZigBee uses the unlicensed ISM bands of 915 MHz and 2.4 GHz in the U.S. and 868 MHz in Europe. All of the chips use direct-sequence spread spectrum (DSSS) for robustness and minimization of multipath. Maximum data rate is 20 kbits/s for the 868-MHz version, 40 kbits/s for the 915-MHz version, and 250 kbits/s for the 2.4-GHz version. Most vendors opt for the 2.4-GHz configuration.

A good example of a typical ZigBee RF chip is Freescale's MC1319x series. The MC13191 is a 2.4-GHz transceiver using DSSS and offset-QPSK (quadrature phase-shift keying) modulation. It's set up so designers can use their own or another proprietary protocol. The MC13192 is similar, but it has a built-in MAC layer that's 802.15.4-compliant. Designers can add their own mesh networking layers above that. The third member of the series, the MC13193, fully complies with ZigBee.

A newer line of Freescale chips, the MC1320x family, puts the transmit/receiver antenna switch on chip. It fully integrates one of Freescale's HCS08 processors on-chip to create a single-chip Zig-Bee solution.

Freescale also has partnered with Millennial Net, which offers mesh networking software and solutions. Millennial Net's Mesh-Scape wireless sensor networking system can run on Freescale's RF transceiver chips, creating a proprietary mesh solution.

MeshScape includes the full mesh networking software as well as a line of node modules. Its 916-MHz nodes and 2.4-GHz nodes work with the software. A MeshGate module aggregates data from the network and sends it on. An End node module captures data in a hybrid star-mesh network, while a Mesh node module is used with sensors and actuators.

The mesh system developed by Ember is a good choice for a full standardized ZigBee implementation. It consists of Ember's EM250 single-chip transceiver/controller—a complete ZigBee system-on-a-chip—as well as all of the software required for a sophisticated mesh. The EM250 has an 802.15.4-compliant, 2.4-GHz transceiver, plus an on-chip 16-bit RISC processor to handle the ZigBee and higher networking layers. Ember's EM260 RF transceiver has the same features as the EM250. It works with Atmel's AVR, TI's MSP430, and other embedded controllers.

The software component of the Ember system, EmberZNet 2.0, is a fully ZigBee-compliant networking stack that can handle a wide range of end applications. Ember's developer kit just about hands your design to you on a platter (Fig. 4). In addition to a developer board, the kit supplies 12 complete node boards that offer designers a chance to test the mesh concept themselves. There's also a Power-over-Ethernet (PoE) injector. On the software side, the kit's full set includes the Ember Studio Network-Management Software, Ember Studio Debug Tools, the EmberNet Stack and API and library, and full documentation.

Lots of other companies produce chips, modules, and software for ZigBee and similar proprietary schemes. For example, Chipcon AS is readying its single-chip-solution CS2420 RF ZigBee transceiver with an integral processor called the CS2430.

Helicomm's IP-Link 1200 modules are ready to use in ZigBee nets. The company also has a 900-MHz module for longer range. Software and developer kits are available as well. Cirronet, another RF modem company, offers a wide range of ZigBee modules, software, and development boards.

WI-FI MESH While ZigBee leads the industry in short-range mesh networking, plain old 802.11 or Wi-Fi can be used in a mesh configuration with the right software. As mentioned earlier, the IEEE 802.11s mesh standard for Wi-Fi is still a few years away. Meanwhile, lots of proprietary systems are available using 802.11 or similar technology. Wi-Fi makes sense for a mesh air interface that needs higher speeds and longer ranges and where power consumption is less of an issue.

Wi-Fi mesh provides low-cost broadband connections to consumers in rural and suburban areas that aren't served by cable TV or DSL lines. By establishing each subscriber as a repeater/router node, Internet service can be provided at low cost over a wide area. Lots of small towns use such a system. Now the industry is looking at bigger fish, though, with one major system already being proposed to offer connectivity all over Philadelphia.

An 802.11 mesh, the Wireless Intelligent Transport Network (WITnet) system from Accton Technology, is a variation of the hybrid star-mesh topology. It provides a way to mesh the APs of existing or new star/P2P 802.11 networks.

Today, each AP needs a connection back to the infrastructure via a T1 line or other link. The Accton system eliminates this expensive link from most of the APs in the mesh. The APs relay data to and from a single wired AP via mesh techniques.

WITnet has all of the properties of a mesh, as it is self-organizing and self-healing. Designers can add new nodes (APs) quickly and simply. It has built-in AES encryption and authentication for security. There's probably no faster or easier way to expand an existing hot-spot network at minimum cost. One additional use is to provide bridging in a wired LAN. On top of that, the system will be compatible with the forthcoming 802.11s mesh standard.

The MEA and Motomesh systems devised by Motorola represent a fast high-end mesh network. Motorola got into the mesh fray when it bought Mesh Networks a few years ago. MEA is the basic Mesh Enabled Architecture that allows 802.11-like nodes and APs in the 2.4-GHz unlicensed band to be meshed. It offers all of the essential features of a mesh and 802.11b systems, and it can be used in mobile applications where 802.11b fails.

MEA uses the patented quadrature-division multiple-access (QDMA) method. Originally developed under a Defense Advanced Research Projects Agency (DARPA) contract for military mesh networking, QDMA is a version of CDMA that doesn't require a basestation like CDMA cell-phone networks. Instead, each node can serve as a repeater/router, and the nodes may be mobile at speeds up to 250 mph. Maximum burst data rate is 6 Mbits/s. Typical sustained rates range from 1 to 2 Mbits/s.

The Motomesh product line incorporates MEA yet goes beyond it to provide very robust mesh networks for public safety organizations. It's a good fit with fire and EMS, police, disasterresponse, and homeland-security organizations. It provides data services as well as streaming video, photos, Geographic Information System data, and access to databases or other information needed in the field by police, construction workers, and others.

At Motomesh's heart lies an access point that combines two standard 2.4-GHz, 802.11b-like radios and two 4.9-GHz, publicservice-band radios that incorporate MEA. One set operates in the standard 2.4-GHz band, the other in the 4.9-GHz band.

These meshed access points can communicate with laptops and other nodes in cars and other vehicles at speeds up to 200 mph (Fig. 5). Its location-based system allows any vehicle to pinpoint itself or any other user within a ±10-m range—without GPS. Even nodes operating inside buildings, tunnels, or downtown "canyons" of tall buildings or traveling at high speeds can be located in about a second.


Accton Technology




Freescale Semiconductor


Millennial Net

www.motorola.com or www.meshnetworks.com

Wireless Industrial Networking Alliance (WINA)

ZigBee Alliance

About the Author

Louis E. Frenzel

Click here to find more of Lou's articles on Electronic Design. 

Sponsored Recommendations


To join the conversation, and become an exclusive member of Electronic Design, create an account today!