Clearing up the mesh about wireless networking topologies: Part 2 - Embedded.com

Clearing up the mesh about wireless networking topologies: Part 2

Having covered the basics of wireless networking in Part 1,as well as the criteria for evaluating the various topologies, it isnow time to look closely at five different network architectures -point-to-multipoint, Zigbee 2007, Wireless Hart, 6LoWPAN, and on-demanddistance vector routing – and assess their strengths and limitations.

Point-to-Multipoint
The point-to-multipoint architecture is also known as a simple star andit is not really a mesh network, but it is also often confused withone.

Keycharacteristics. Such networks tend to use the modern airInterfaces of either FrequencyHopping Spread Spectrum or DirectSequence Spread Spectrum (802.15.4). They need to be staticallyconfigured for PAN ID,routes, and security.

It is important to note that all of the nodes can see other nodesand that they need to be told who to talk to. Security tends to bepair-wise for both the encryption and key. End points may go to sleepor stay awake, but the central router is always awake.

NetworkArchitecture. Figure 4 below illustrates a typical topology. All nodes are on the same channel (orhop to the same channel). Bandwidth / throughput is limited bysimultaneous data at concentration point. Collisions happen with lotsof traffic or lots of nodes.

Figure4. A typical point to multipoint wireless configuration. Green endpoints see other end points but are told only to talk to the orangecentral point.

Strengths. The beauty of the basic non-mesh point-to-multipoint network issimplicity. Communication, unless traffic is very heavy, is relativelydeterministic since there are no hops and minimal or managedcollisions.

It also allows for maximum throughput because there is no addedrouting and no added route discovery. Finally, it is easy to understandand easy to manage. Because of the simplicity, it also tends to be thelowest cost for its specific size and function.

Limitations. Unfortunately, the simplicity described in the strengths also drives anumber of limitations. The networks will tend to be small. Largenetworks only work if polled from a central point.

This requires very specific message management. There are alsosingle points of failure and no ways to route around changingconditions. The network follows the belief that if it worked the firsttime, it will work forever so you must be sure of good RF conditions.

ZigBee 2007
ZigBeeis built on top of 802.15.4 using DSSS in 2.4 GHz. End points sleep,routers don't sleep and a coordinator is needed to start the networkand to allow points to join the network.

Keycharacteristics. ZigBee has had three different versions of thestandard ” 2004, 2006 and 2007. ZigBee 2004 is no longer used andZigBee 2006 had significant limitations. ZigBee 2007 includes keyfeatures for frequency agility, message fragmentation and enhancedsecurity associated with key management.

The routing of messages follows the previously describedCluster-Tree methodology where routes to all points are maintained ateach cluster. This allows a very short routing time, but requires lotsof routes. Discovery of routes uses the AODV algorithm where paths areexplored between clusters.

NetworkArchitecture. The network consists of three specific types ofpoints. The ZigBee Coordinator (ZC) has one required for each networkand it initiates network formation. The coordinator may act as a routeronce the network is formed.

The ZigBee Router (ZR) is actually an optional network component,although a network without routers becomes a point-to-multipointnetwork described earlier.

The router participates in multi-hop routing of messages. Finally,the ZigBee End Device (ZED) does not allow association and does notparticipate in routing. As such it is often referred to as a childbecause it doesn't really have any responsibilities. Figure 5 below illustrates a Zigbee2007 network.

Figure5. A typical Zigbee 2007 wireless network configuration.

Strengths. Enddevices are very low power because they are subservient to parentalrouters. Cluster-Tree routing provides quick knowledge of routes andthereby efficient routing.

With ZigBee 2007, frequency agility switches from problem channelsautomatically in a sort of on demand frequency hopping. Long messagesare allowed with message fragmentation support and security is flexiblewith support of separated keys. Finally, the network can scale to bevery large.

Limitations. The biggest limitation tends to be in terms of power in the routers.Routers must be powered ” they can never go to sleep. In addition, thebenefit of Cluster-Tree routing also means that network changes requirea lot of route discovery traffic. Heavy traffic volume means lots ofcollisions and potential message loss. Finally, a coordinator is neededto start and manage the network, so if the coordinator goes down, noone can join and the network can't start.

Wireless HART
Wireless HART uses the Time Synchronized Mesh Protocol (TSMP) createdby Dust Networks.Unlike other networks, the time based system uses TDMA (Time Slots) foran access method.

KeyCharacteristics. The network is optimized for low power and allnodes can be sleeping routers and every node is a router. A gateway isrequired to keep the network synchronized due to the critical timesynchronization of sleeping and waking functions. Like ZigBee, it isbuilt on top of 802.15.4 DSSS, but it adds a more deliberate frequencyhopping algorithm. Security includes encryption and authentication.

Network Architecture. Figure 6below illustrates a typical network topology. Note that all thenodes are routers. The illustrated routes change dynamically based onvisibility within specific time slots as it hops through the differentDSSS channels.

The relationship between any two nodes is negotiated to be in aspecific time slot, thereby minimizing the probability of anycollisions. When sleeping, nodes awaken during their time slot andlisten to see if there is any traffic. Clocks are kept synchronized bythe gateway.

Figure6. A Wireless Hart Time synchronized mesh network configuration.

Strengths. Everynode is a router at very low power consumption. Most of the time isspent listening. Since transmissions occur only within the allocatedtime slot, retransmissions are minimized.

Communications are very reliable with every message acknowledged.Networks are able to scale to moderate level or around 1000 nodes.Frequency hopping minimizes the probability of interference. Securityincludes encryption and appropriate authentication.

Limitations. Because of the time slot approach, latency is long andnon-deterministic. It takes a network a while to form and all of thenodes to negotiate their individual time slots. Because communicationsis slotted, the available 802.15.4 bandwidth is split up, meaning thatthroughput is minimized for bursty traffic.

A powered gateway (coordinator) is required for the network to stayfunctioning opening up a single point of failure if the gateway isunavailable for an extended period of time. Finally, the radios arevery expensive compared to the other available solutions.

6LoWPAN
6LoWPAN is a distortedacronym for IPv6 over low power wireless personal area networks.Presently it is a proposed standard based on the IETF RCF 4944. It isdesigned to be used over 802.15.4 chips and radios.

KeyCharacteristics. Unlike traditional IPv6, 6LoWPANdeals with packet size incompatibilities in message transport (128bytes vs. MTU of 1280 bytes in IPv6) and it is designed for a smallmemory footprint system. Today it is a point-to-multipoint architectureand it is proposed to be augmented with a mesh routing scheme.

NetworkArchitecture. Figure 7 below illustratesan example network topology. Note that for now it is onlypoint-to-multipoint. Unlike the other networks discussed, the figureshows an end to end IP based link from a host computer to an enddevice.

In this case it is illustrated by a meter. The end device isdirectly addressable by the host computer on the far end of thenetwork. The interworking function provided in the pictured boxprovides a transport change and re-packetization at the data-linklevel.

Figure7. A 6LoWPAN IPv6 over wireless network configuration.

Strengths. The most powerful strength is that 6LoWPAN is able to take advantage ofthe existing TCP/IP suite of internet protocols, all of which are wellunderstood due to the proliferation of the internet. Hence it is ableto capitalize on existing protocols, existing quality of service andsecurity framework supported by the IETF. Hence, it enables seamlessrouting of message payloads.

Limitations. This system is still very new and is only a proposedstandard. Because it is officially in the public review stage, it willmost likely undergo a number of changes.

In fact, the mesh routing working groups are still being formedmeaning that wide scale adoption is still a few years away. As such,interoperability is a nice concept that has not been proven yet.Finally, because it is still new, it has not yet been ported to a largegroup of chipsets.

Ad hoc On-Demand distancevector routing (AODV)
Like its sibling Wireless HART, AODV (including the variant used in theDigi Mesh)is designed to meet the very low power sensor networks where batterypowered routers are required.

KeyCharacteristics. Our AODV variant is available in multiplefrequency bands, 2.4 GHz DSSS and 900 MHz FHSS and does not rely on afull 802.15.4 implementation as it has some of these functionsinternal.

For both message routing and discovery, it uses a variant of AODV,meaning that routing tables are built only for needed destinations,leaving it to be referred to as a peer-to-peer mesh instead of acluster-tree.

All nodes are viewed as equal participants meaning that they are allrouters and they can all sleep. Channel access is a sort of timesynchronized CSMA method, enabling bursty traffic, but the benefits offew collisions. It has a full security suite.

NetworkArchitecture. Figure 8 below illustratesa typical ad hoc network topology. Unlike the Cluster-Tree methoddescribed in ZigBee, routes are only determined on an as needed basis.This means that routes that are never used never get routing tableentries and routes that are used frequently are continuously updated,optimizing their efficiency.

Figure8. The Digi Mesh AODV wireless network topology.

One of the other keys to note about the topology is that there is nocoordinator or gateway function. Time synchronization is accomplishedthrough a nomination and election process, enabling the network tooperate autonomously.

RoutingMethodology. Figure 9 below shows the process of how routing failures are handled. The first showsthe initial network configuration where a route has been establishedfrom one point to another.

The second illustrates a failure where one of the nodes has beenremoved for an unknown reason, removing relationships in the center ofthe route. Finally, the last figure shows how this route isreconstituted using a path that didn't previously exist. Therelationships were there, but they had never been used, but were newlydiscovered using AODV after the failure.

Figure9. Here's how routing failures are handled with an AODV wireless meshnetwork configuration.

Strengths. Every node is a router at very low power consumption. Further, becauseevery message is acknowledged and routes are determined on an as neededbasis, the network is not overwhelmed with unnecessary discoverytraffic ” very important if the routers are battery powered andsleeping.

Efficient route discovery and routing means that the network onlylearns routes that actually get used (AODV). Frequency agility issupported and security meets the requirements of both encryption andauthentication. Reliability is projected at 99.99%. Finally, the systemsupports larger Payloads with support for message fragmentation.

Limitations. Unfortunately, efficient power management means latency is long andnon-deterministic. Even though throughput is not limited by time slots,it is still limited depending on loading and discoveries. The networkcan scale to a moderate size of around 500+ nodes and can be very largeif traffic is light and message flow doesn't change much.

Comparison of the alternatives
Using the criteria defined earlier in Part 1 in this series, Table 1 below illustrates my bestattempt at evaluating the different network approaches.

It is important to note they all do very well in security in thatthey have well defined encryption, authentication and authorizationschemes. ZigBee and 6LoWPAN get a slight nod here only in that theirkey systems should be easier to implement and a bit more flexible.

Table1. Comparing the wireless mesh alternatives

With respect to reliability, Point-to-Multipoint takes the biggesthit because it inherently has a single point of failure. Some schemesmay have frequency agility options while others do not.

Prior to the 2007 standard, Zigbee has a weakness in the frequencyagility area; this is fixed in the 2007 standard along with addingsupport for message fragmentation.

The others are similar – Wireless HART is designed to never lose amessage so it gets the nod here while 6LoWPAN does well on theassumption that the existing TCP/IP protocol suite has class of servicebuilt in. While the AODV-based Digi Mesh has a similar approach toWireless HART, it is still somewhat unproven in large deployments.

Power management will no doubt be hotly debated. The nod was givento Wireless HART and our AODV variant because they both define systemswhere all nodes in the network, including routers, can sleep.

Even though sleeping Zigbee end devices are most efficient when itcomes to power, the fact that routers can't sleep bumped the ratingdown. Until 6LoWPAN settles on a mesh and power management strategy,the rating will remain unknown.

The scalability rating follows directly from the question of how bigcan the network get and still function. This is where the Zigbee 2007Pro stack shines. The Cluster-Tree architecture creates a hierarchywhich enables scalability.

Digi's AODV variant and Wireless HART scale well; particularly ifmost communication is kept local ” however, the networks can tend toget very slow when they get too big. Finally, point-to-multipoint hasan obvious limitation in the number of nodes that can be attached toone central point.

The best data mover is no doubt the simplest system – namelypoint-to-multipoint. The simple network design means that focus can bemade on short, deterministic latency and high data throughput.

There is a direct trade-off here with power. Wireless HART and DigiMesh rate lower here because they are focused on minimizing power andmaximizing reliability ” this naturally leads to less deterministiclatency and lower throughput.

I recognize of course, that as a network gets bigger, these twonetworks will actually do better; however, this is represented in thehigh scalability ratings for these networks. Zigbee fits in the middlehere because the backbone of powered routers can move data veryefficiently ” but can get stuck if too many route discoveries areneeded.

<>Cost may end up with the most debate. The ratings here were basedprimarily on the view of the cost of available chip set solutions underthe assumption that the right architecture is chosen for the right job.If not, then the cost ratings go out the window. For example, trying todeploy a Zigbee solution where battery powered routers are desiredmeans infrastructure costs will skyrocket. So given this caveat,point-to-multipoint, Zigbee and Digi Mesh have common costs becausethey all use similar chipsets. 6LoWPAN is somewhat unknown ” dependingon resource requirements.

The assumption is that similar to current chipsets can be usedwithout substantial feature degradation. Wireless HART has a low ratingpredominantly because the limited number of suppliers has kept chipsetprices 5X to 10X other solutions and customers have not demanded lowercosts due to primary use on expensive assets in process controlenvironments. This will most likely change as more competitors enterthe market.

Conclusion
We have traced the architectures of wireless mesh networks and therespective architectural trade-offs. Each of the Wireless MeshArchitectures has respective benefits as they optimize on differentcomponents. There is not a one size fits all approach as throughput istraded off against reliability and power consumption.

Hence, it is important to match the needs of the application to thecapabilities of the network. Further, it is important not to settle forthe wrong network because of fad or hype in the market place.

No doubt many of the conclusions here will be hotly contested bydifferent network architectural advocates. This is always true wherethere are shades of gray in evaluation of different criteria. Forexample, had this article been done a year ago, the results will havelooked very different ” as they will look different a year from now.

To read Part 1, go to The basics of wireless mesh networkingtopologies .

Joel Young has more than 15 yearsof experience in developing and managing data and voice communications.Mr. Young joined Digi International as VicePresident of Engineering in June 2000 and is currently the VicePresident of Research and Development and Chief Technical Officer ofDigi. In his current role, Mr. Young is responsible for research anddevelopment of all of Digi's core products.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.