Design Article
Solutions arrive for ad hoc nets
Peter Stanforth, CTO, MeshNetworks Inc., Maitland, Fla.
2/14/2003 1:11 PM EST
A peer-to-peer wireless system offers considerable advantages over more-traditional wireless-network architectures such as cellular or wireless local-area networks. However, it also poses different technical challenges. An ad hoc wireless network is a collection of mobile terminals-handheld devices, mobile phones, automotive telematics systems-that communicate directly with each other without the aid of established infrastructure. Through multihop routing techniques, the terminals act as routers/relays for one another and extend the range and coverage of communications links between parties.
Some industry cynics claim that ad hoc wireless networks cannot scale, do not provide high throughput and are too complex to manage, making the technology commercially unfeasible. Contrary to that viewpoint, a growing number of commercial companies are developing and deploying products based on ad hoc peer-to-peer wireless technology. This shows that these issues can be successfully resolved, as some already are. In addition to the companies deploying the technology for commercial use today, the Internet Engineering Task Force, through its mobile Internet Protocol and mobile ad hoc networking initiatives, is actively striving to standardize routing protocols and architecture solutions. Performance of the technology in real life, however, will be dictated by the specific system architecture and mechanisms used to balance engineering trade-offs in the most efficient manner.
The fundamental benefit of a mesh topology is its ability to significantly improve the capability of any RF modulation scheme by:
- Improving frequency reuse over other architectures;
- Requiring less power to achieve high data rates;
- Creating self-forming and self-healing network routes;
A meshed architecture improves frequency reuse via a combination of power control and hopping technology. Large distances between senders and receivers are covered by leveraging intermediate nodes (either infrastructure or other users) as low-power relays. Nodes "whisper" to each other and create small zones of interference that are only as wide as a typical pico or micro cell along the path of the transmission. Since the zone of interference is so narrow, spatial reuse of frequencies is very high. This same technique of hopping through network nodes also dramatically decreases the total end-to-end power expenditure for any given end-to-end data rate.
Reduction of power is achieved by choosing to transmit over multiple short distances vs. a single long path. This translates into either a lower transmit power or a higher data transmission speed over a given distance. Either one will decrease the total end-to-end energy needed to transmit a packet of data. This can be demonstrated simply with an 802.11b network by placing a "hopping" node half way between a transmitting node and the receiver.
If the transmitting node is able to maintain only a 1-Mbit data rate directly because of path loss or interference, the 7 dB of gain achieved by halving the transmit range (via hopping) typically results in the data rate improving to 11 Mbits. By using lower power and a hopping scheme, the devices achieve higher system throughput and create less interference in their environment.
Because pure ad hoc networking requires no infrastructure, the clients must cooperate to organize into a network and resolve contention for the available bandwidth among themselves. These tasks become more complex as the number of nodes grows or if the relationship between nodes changes rapidly as in a mobile network. Without specific action being taken, the network will grind to a halt as the nodes consume most of the network bandwidth . The challenges lie in creating low-overhead mediaaccess and routing protocols.
Most ad hoc wireless networks use some form of carrier-sense multiple-access (CSMA) random-access scheme, similar to the protocols defined by Apple for Appletalk and made wireless by Proxim. These protocols have to be extended to deal with hidden-node and exposed-node problems. When node E tries to communicate with D it may not hear a concurrent attempt by C to talk to D. C is a hidden node to E. In contrast, if node B wants to talk to node A, but hears a communication from C to D, B may decide that the channel is busy and not attempt to send to A. B becomes an exposed node to C.
These collisions, caused by the nonsynchronized nature of the protocol, create inefficiencies. If not managed, they can be so damaging that large networks choke.
Ad hoc routing can be divided into proactive and reactive protocols. A proactive protocol keeps track of all the nodes in its neighborhood, having total knowledge of all possible routes at all times. Reactive protocols do not maintain routing information proactively and find routes only when they need them.
Even though the technology to make this all work is complex, there are many new and existing applications where ad hoc peer-to-peer wireless networks are being deployed.
The technology's characteristic lack of infrastructure is very appealing to the military and to public-safety first-responders. Machine-to-machine communications include sensor networks, process control and automated dispatch. Ultimately, multihopping networks can exploit non-line-of-sight capabilities. Wireless mesh networking clearly has the potential to deliver the true wireless Internet that will supplement or replace today's cellular systems.
See related chart
|
|
|
|
|
