WirelessBR

UMA AULA DE MESH TECHNOLOGY!

Venha assistir ...

Community Mesh Developments
Dailywireless
Thursday, 2006 February 02 @ 10:01:09 PST by Sam Churchill (153 reads)

ABI Research, in a report called "Wireless Mesh Networking", examines the trends for metro-scale and campus-scale wireless mesh networking technology on a worldwide basis.

Equipment vendors profiled include; BelAir Networks, Cisco Systems, Firetide, MeshDynamics, Motorola, Nortel Networks, PacketHop, Rajant, RoamAD, SkyPilot and Strix Systems.

Wireless Newsfactor describes the different architectures:

(1) Tropos' MetroMesh OS with PWRP can scale to thousands of routers in a mesh network without exceeding 5 percent of available bandwidth. It achieves this by measuring throughput 5 times a second on all available paths back to the wired network. Then, each router selects the path with the highest throughput.

(2) Firetide's EtherDirect enables the mesh to operate as several small local clusters of nodes. Each cluster can support about 30 nodes and connects to other clusters.

(3) MeshDynamics' intelligent Structured Mesh nodes are configured with one 802.11b/g radio for client access and two 802.11a radios for backhaul.

(4) The BelAir 200 features a plethora of backbone radio options, providing antenna selection, to switch the signal path in the appropriate direction to eliminate the need for the installer to manually point the antennas.

Strix configures its network with up to six radios per node.

SkyPilot has a high-gain, 8-antenna array and offers 4.9 GHz public safety option.

Cisco has a dual radio system coupled with their popular controllers and promotes mesh standards using the Lightweight Access Point Protocol.

ABI notes that BelAir formed a cable Technical Advisory Board (see: DailyWireless Cable Gets Meshed). Tropos is also working with cable equipment supplier Scientific Atlanta, which planned to resell Tropos equipment along with its products.

But Cisco recently bought Scientific Atlanta and Cisco has a mesh product of its own which features the Lightweight Access Point Protocol (LWAPP) for standardized communications protocol between access points and WLAN systems.

Motorola's MotoMesh consists of three devices: an Intelligent Access Point; wireless routers; and the optional MEA radios themselves. Each IAP contains four radios: two each on the 2.4-GHz and 4.9-GHz bands, permitting communications in the unlicensed and licensed spectrum. One of the radios in each spectrum is configured to use Wi-Fi, while one uses the MEA protocol.

Motorola said Monday, it has taken its MotoMesh mesh network out of trials and placed it into general production for metropolitan areas.

Taipei's Huge Cloud uses some 3,300 Nortel mesh access points.

ABI says a different picture is emerging, in Europe. With 3G deployments more advanced in Europe, mesh community networking trails there compared to North America or Asia-Pacific.

More ...

Inside Wireless Mesh Networks
NewsFactor Network
February 2, 2006 7:32AM

Resiliency is a key attribute of mesh networks. A mesh network supports multiple paths among network nodes, hence there is no single point of failure, and nodes can be added to increase redundancy. Mesh networks also can load-balance traffic to prevent congestion and reroute traffic around congested nodes.

Mesh networks are a hot new trend in wireless. In fact, mesh technology is the next step in the evolution of wireless architecture, promising greater flexibility, reliability and performance over conventional Wi-Fi wireless LANs (WLANs).
Unlike WLANs that have a wired connection between access points and switches or a wired LAN, in wireless mesh networks (WMNs), all communication among mesh network nodes, and between nodes and clients, is over a radio link instead of cable.

Some WMN vendors additionally support connectivity between their mesh networks and an enterprise network via either radio (802.11a/b/g) or direct Ethernet (100BaseT) cable connection for users or service providers who want to interconnect the two networks.

The major benefit of this emerging technology is that it enables anyone with a wireless adapter card in their laptop, tablet or PDA to communicate on a peer-to-peer basis, in addition to accessing a WLAN. The key advantages of wireless mesh networks include ease of installation, no cable costs, automatic connection among all nodes, network flexibility, automatic discovery of newly added nodes, redundancy and self-healing reliability. According to a recent study by ABI Research, the use of mesh networking technology for large-scale wireless deployments could easily grow by a factor of 10 over the next five years.

Wireless mesh networks can be deployed outdoors or indoors, in the home or small office. The deployment of outdoor WMNs for a metropolitan area, such as citywide or community-wide "hot zones," or large "hot spots," e.g., corporate or university campuses, will likely contribute to the anticipated substantial market growth.

Wireless mesh network players are growing in number and now include more than a half-dozen privately-held startups, including Co Co Communications, BelAir Networks, Firetide, Mesh Dynamics, PacketHop, Sky Pilot, Strix Systems and Tropos Networks, each of which provides its own hardware/software solution.

In addition, legacy network vendors that currently offer WMN products include Motorola MeshNetworks, Nortel, and most recently, Cisco. The entrance of Cisco into WMNs further validates the industry and bolsters growth predictions.

There are two basic types of WMN architectures: fixed and mobile. Fixed mesh networks contain network nodes deployed throughout an indoor or outdoor area or zone specified for wireless connectivity. Each physical node houses routers, radios and one or more antennas. Outdoor nodes are mounted on utility poles or buildings throughout a campus or metropolitan area to form a wireless infrastructure referred to as a "hot spot" or zone. Some WMN vendors support both indoor and outdoor fixed WMNs with seamless connection between the two.

In contrast, mobile WMNs are tuned for high mobility and enable the creation of autonomous mobile networks composed of off-the-shelf devices, each containing an 802.11 wireless radio. Mobile mesh networks can be deployed instantaneously anywhere and provide broadband coverage wherever users are. Mobile networks grow in capacity, range and redundant connections as more users join. Additional client nodes fill in gaps in the wireless coverage.

Mobile WMNs have been used by the military to establish temporary networks under battlefield conditions. They are also applicable to ad hoc applications by first responders, such as police, firefighters and EMS crews because they can be immediately established and then terminated when a mission is completed.

All the vendors mentioned above, except Co Co Communications and PacketHop, provide fixed WMNs. Motorola MeshNetworks and Nortel provides both fixed and mobile mesh networks, and each of these two large vendors has partnered with PacketHop. PacketHop and Co Co Communications each offer software that creates an instant mobile network anywhere, on the fly. Both companies are targeting first responders.

The Upside

Mesh technology opens up new applications and offers several benefits over conventional Wi-Fi WLANs. A WMN is easier to install and is less expensive than wired networks, since it uses radio signals instead of cable to connect nodes.

Resiliency is a key attribute of WMNs. A mesh network supports multiple paths among network nodes, hence there is no single point of failure, and nodes can be added to increase redundancy. Mesh networks can load-balance traffic to prevent congestion and reroute traffic around congested nodes. A mesh network distributes intelligence throughout the network instead of centralizing it, as is done in switch-based WLANs. WMNs can also reduce or eliminate T1, DSL and other service costs for backhaul traffic.

Another key benefit of mesh networks is their ability to circumvent large objects, such as buildings, that block the signal path; signals are simply re-routed through alternate nodes that bypass the building or object. Nodes can also be added to cover dead zones or to compensate for blocked signal paths.

The Downside

Mesh networks also have their down side. WMNs can suffer from bandwidth degradation, radio interference, and per-hop latency as networks grow. But increasing the size of a mesh network doesn't automatically degrade performance. In some cases, performance is actually improved by the added capacity and load balancing.

Bandwidth degradation and latency occur when traffic travels between WMN nodes. For each hop, bandwidth is cut in half (bandwidth = 1/n where n = the number of hops) and latency increases (each effect occurs for technical reasons explained below). For example, a hop count of four would reduce throughput to 25 percent.

A worst-case scenario-and also, unfortunately, the most realistic one-for bandwidth degradation is where the bandwidth degrades by (1/2) n-1. This is the most realistic case because in most mesh networks, each node is surrounded by several neighboring nodes that cause radio interference. In this case, throughput would be reduced to 13 percent for a hop count of four. So the key to better performance is sound network planning-at least in fixed networks, where such control can be fully exercised. Network planners must try to minimize excessive hop counts and to balance loads.

Radio frequency interference (RFI) can considerably affect the devices on a wireless mesh network, causing transmission errors and degraded network performance. RFI can be radiated from a whole gamut of sources; e.g., other electronic devices, cordless phones, wireless security cameras, microwave ovens, computers, etc. RFI can be continuous, temporary or intermittent, depending on the activity of the RFI source.

Many WMN devices have omnidirectional antennas, which extend WMN connectivity but also can be another limiting factor. Since omnidirectional antennas propagate a signal in a 360degree pattern, the network suffers from RF interference and collisions as the number of users and nodal transmitters on the network increase.

An omnidirectional antenna also limits the coverage range of each mesh node, because its energy is dispersed equally in all directions instead of being focused in one direction. In contrast, directional antennas, as the name suggests, concentrate all their energy in one direction. High gain antennas extend signal range, allowing nodes to be placed further apart to require fewer nodes within a network, helping to achieve the network planning goals described above.

Fixed WMN vendors include one or more antennas in their network nodes. Some, such as BelAir Networks, provide antenna selection, to switch the signal path in the appropriate direction to eliminate the need for the installer to manually point the antennas. The use of multiple frequencies and channels within a mesh network can preclude interference. Firetide's MeshBridge feature does just that for each mesh segment, to optimize a segment to local operating conditions.

Network latency and jitter affect wireless mesh networks that support data, voice, and video traffic. Voice and video performance degrade as latency increases and jitter becomes extensive, so hop counts must be kept down in order to minimize latency.

Both mobile and fixed WMN vendors use special routing algorithms and protocols to route packets. Legacy routing protocols, such as link state or distance vector protocols and the Spanning Tree Protocol can't be used in WMN, because their overhead traffic consumes too much bandwidth, degrading performance.

Tropos Networks developed its own proprietary routing protocol it calls Predictive Wireless Routing Protocol (PWRP), which it designed for a dynamic mesh environment. It optimizes path selection based on end-toend throughput optimization. PWRP minimizes packet retransmissions throughout the mesh, ensuring high capacity and low latency.

In addition, Tropos claims its MetroMesh OS software with PWRP can scale to thousands of its WMN routers in a mesh network without exceeding 5 percent of available bandwidth. It achieves this by measuring throughput 5 times a second on all available paths back to the wired network. Then, each router selects the path with the highest throughput. This method eliminates traditional routing tables with predetermined paths. PacketHop's network solution also uses network overhead of just 1-5 percent, the precise amount depending on mesh size. Performance is substantially improved with such low overhead.

While most WMN vendors use proprietary protocols to route packets over their network, Strix Systems believes that routing overhead ultimately limits network scalability and has implemented Layer 2 switching to accommodate voice applications and fast-roaming handoff for minimum latency and overhead and improved multihop performance. Each radio module within a Strix node contains a switch, which establishes the optimum path through the network as determined by its switching parameters.

Scalability and Performance

The scalability and performance of WMNs rely heavily on the network's architecture to overcome the limitations previously cited. For example, early-generation mesh networks had limited scalability and suffered performance degradation over multiple hops. That's because each node contains a single radio, operates in half-duplex mode, and shares the same radio frequency. This is referred to as a Basic Service Set (BSS).

All radios on the same channel must remain silent until the packet completes its total hops within the same BSS. Each sequential hop reduces bandwidth by one-half of that at the previous node, because the node must handle the packet twice-receiving it and then sending it on.

In addition, waiting for a packet to complete its total number of hops introduces substantial delay; latency of up to 50 milliseconds per hop can be introduced, which at four hops is 200 milliseconds. This much delay precludes delay-intolerant applications such as voice and real-time video.

As noted above, excessive hop counts can be minimized through good network architecture design. For example, Firetide's EtherDirect feature enables the mesh to operate as several small local clusters of nodes across wider areas. Each cluster can support about 30 nodes and connects to other clusters with a high-speed wired connection. Firetide's Multimesh Management can support a very large number of clusters, creating a single mesh environment of up to 1,000 nodes.

While single-radio networks tend to be the least expensive, scalability and performance can be greatly improved via more-sophisticated (and more expensive) multiple-radio mesh networks. This architecture provides a "backbone," called the backhaul network, which interconnects all nodes and handles traffic between nodes. Separate radios within each node, tuned to different frequencies, reduce backhaul interference and congestion. Client access is supported on separate 802.11a/b/g links provided by each node. Vendors refer to this type of network as a structured mesh network, since it differs from "unstructured," ad hoc mesh networks.

In two-radio mesh networks, each node contains one radio for backhaul, while a second radio acts as an Access Point (AP). In other words, one of the node's radios "talks" to the client, and the other radio "talks" to peer nodes.

Although the dual-radio architecture improves performance and scalability, even greater performance and scalability can be delivered with a more complex multi-radio architecture, which uses three or more radios per node. A three-radio node adds a second radio for the backhaul portion of the network; having two radios per node for backhaul enables a node to operate in full-duplex mode, so it can transmit and receive backhaul traffic concurrently.

In practice, mesh network architecture may combine single-, dual, and multi-radio nodes within a network to address specific network requirements and to maintain the lowest cost. Because single-radio network segments have limited scalability, they are limited to fewer nodes to reduce hop counts.

Examples of multi-radio (i.e., three or more radios) systems include MeshDynamics' intelligent StructuredMesh nodes, which are configured with one 802.11b/g radio for client access and two 802.11a radios for backhaul. Strix configures its network with up to six radios per node. Core nodes contain four to six radios, while access nodes contain two or four radios, and edge nodes have one or two radios since they do not relay any traffic.

Another vendor, BelAir Networks also supports single-, dual- and multi-radio mesh networks. It typically uses single- and dual-radio nodes at the network edge with multi-radio nodes at the core of its backhaul network. BelAir limits its single-radio mesh clusters to eight nodes per cluster, which typically results in a maximum hop count of three hops per cluster.

While mesh network vendors claim their networks do not have hop count limits, they design their WMNs for low hop counts of typically three or four maximum. This is especially true for vendors such as SkyPilot, which reports that many of its implementations support customers using voice over IP, where end-to-end latency is a critical factor. SkyPilot reports that each hop adds 8 to 10 milliseconds of round-trip latency, so it limits its hop counts to four or five hops to maintain latency under 50 milliseconds.

Absent the requirements of an application such as VOIP, SkyPilot can scale much higher. The company says it has production networks that scale up to more than 350 mesh nodes supporting up to 70 client devices per node. It claims it can scale up to 1,000 mesh nodes.

Large-scale WMNs provide metropolitan coverage. For example, Tropos Networks has deployed its MetroMesh network for continuous coverage in several U.S. cities, including Oklahoma City (640 square miles), Corpus Christi, TX (147 square miles), and Chaska, MN (18 square miles). Strix Systems reported a newly created network in Tempe, AZ that supports voice/video/data operation over a 40-square mile area. Mortel claims to have the largest metropolitan network in the world, with 10,000 APs covering 270 square kilometers in Taipei, the capital of Taiwan.

Network Reliability

High reliability is the hallmark of wireless mesh networks due to their fault-tolerant architecture. Self-healing failover capabilities automatically sense a path outage and reroute traffic. Some mesh network vendors provide backup radios and antennas for backhaul and access portions of the network for even greater reliability.

Mesh networks also support traffic load balancing over multiple backhaul links, in response to increasingly heavy traffic. Power supply redundancy and battery power backup are also necessary for highest resiliency. Networks designed for outdoors provide rugged construction and weatherproof housing, in addition to features such as lightning protection, electrostatic discharge protection and internal heaters for low-temperature operation in cold environments.

Network Security

WMNs provide the same security features as WiFi WLANs. Backhaul networks must have the greatest protection and are typically encrypted, because they support all traffic for the entire network. SkyPilot, for example, provides 128-bit AES (Advanced Encryption Standard) on each wireless node in its backhaul network and employs MD-5 certificates to ensure authorization.

WMN vendors support standard client authorization measures on their Wi-Fi access points including 802.1x and RADIUS servers . Vendors also support a variety of encryption techniques, such as WEP, WPA, EAP-TTLS, EAP-PEAP, TKIP (Temporal Key Integrity), 802.11i etc., to support mobile client requirements.

Standards and Interoperability

WMNs provide standards-based network access using 802.11 on the network access radios within the mesh AP, to communicate with user's 802.11 cards in laptops and handheld devices. Most WMNs support 802.11b/g. But because commercially-available wireless mesh networks are relatively new, there is no IEEE-based standard yet for creating the mesh among the nodes.

Today's enterprise mesh networks are based on vendor-proprietary mesh algorithms and routing protocols, and do not support interoperability between competitive products. More than 70 protocols have been introduced for routing data across mesh networks.

The IEEE 802.11 Working Group has created Task Group S (TGs) to develop a mesh network standard. At the IEEE 802.11 meeting in San Francisco in November 2005, 15 proposals were submitted for a WMN standard. TGs hopes to have a completed draft standard in 12 to 18 months, which means the 802.11 s standard could be ratified before 2008 if all goes smoothly.

Interoperability will require thorough testing and certification by a consortium such as the WiFi Forum. Some vendors, such as BelAir, SkyPilot, Strix and Tropos plan interoperability testing, though such testing has not yet begun. BelAir has conducted interoperability tests with client devices, such as Wi-Fi phones including SpectraLink and Vocera; and with gateways including Nomadix and IP3, Ponto and Airpath OSS products and PacketHop's mobile mesh. Firetide is developing a program to certify interoperability on a Firetide mesh network.

Conclusion

WMNs are available now for indoor and outdoor applications. Many have been deployed over the past two years and many more are pending. Outdoor metro and campus networks will provide the greatest coverage and revenue.

Network vendors have overcome the basic limitations of mesh networks with sophisticated, multi-radio network architectures, and advanced technology that enables robust large-scale, high-performance networks. However, WMNs are proprietary and a mesh standard will not be ratified anytime soon, so WMNs produced by different vendors cannot be interconnected.

Since there is great diversity among WMNs, prospective users should carefully evaluate vendor products for applicability to their specific networking applications and requirements, visit WMN installations and talk to users before investing heavily in this new proprietary technology.