Why Use A Mesh Network?

Can you guys who are using mesh networks tell me why its beter than point to point and point to multipoint with multiple back hauls. I can see how it would benefit a small town with live feeds in police cars. What are the other benefits?


In theory, a mesh network is a "self healing" communications method in a technology that is typically plagued with unplanned issues. If you have a radio mounted on a street pole and a drunk driver wipes it out, you have to replace it to get your communications back up. Maybe a radio gets blocked by a temporary obstacle. In a well defined and installed mesh system, there would be an alternate path and the connection would heal without intervention.

Good theory.

The high-level tradeoffs are:

  • PTP and PTMP for line of site connections to a small number of fixed cameras.
  • Mesh for cameras that are moving or beyond line of sight.

Mesh is far more expensive and difficult to setup so it should really only be used if PTP or PTMP can't be feasibly implemented.

Mesh used to be popular in surveillance, when Firetide had a ton of marketing money but with the growth of super low cost PTP and PTMP gear, that's changed.

Related:

I agree with John here. The MESH hardware manufacturers were targeting surveillance integrators with very little wireless experience. They pushed their solution as a "set it and forget it" technology that could be deployed with a modest level of training and experience. The software was supposed to do all the difficult work.

The reality is that the deployment in a city surveillance environment is difficult to troubleshoot when nodes drop off line. When the system doesn't "self heal", what do you do? That is when most inetegrators got into the weeds and the simple suddenly became a nightmare.

PTP and PTMP networks are easier to design and maintain, and as John pointed out, are much more cost effective. There will always be applications for MESH technologies, but I wouldn't buy into the hype that they are idiot-proof and self-configuring.

Mesh networks were the big buzz word (like "cloud") in the middle of last decade. Companies such as Belaire, Tropos, Skypilot, Motorolla (MeshNetworks), MeshDynamics, and others were fighting to provide city networks. Later, it was discovered that this did not work unless there was a bridge devices to move inside buildings (I believe this is where Ruckus go their start).

Once alot of companies and cities lost their shirts (Metrofi, Philidelphia, etc), it was determined that this technology is best suited for public safety. (I installed a citywide SkyPilot 4.9 GHz in 2006 which is still running and being used today). Problems come down to physics, though. In my case, the SkyPilot could do maybe 10-12 Mbps from the Gateway, and 2-4 Mbps from a one hop a way Extender, and 1-2 Mbps two hops away from and Extender, all the while increasing jitter and bandwidth instability (H.264/MPEG4 hates this). A total network had 20 Mbps capacity (802.11a) using a simplex single frequency design. PTZ cameras were sluggish and inconsistent. (Nowadays, SkyPilot is still in existence, as it was bought by Trilliant for smart meters).

With Fluidmesh and Firetide, newer modulations and schemes are developed to increase bandwidth, in addition to multiple frequency designs. I have to say that I have never used these, but the physics are the same. Minimal capacity on systems, when you consider HD cameras and their bandwidth/latency/jitter requirements. There may be several "backhaul" and "client" frequencies used, but the problem becomes "omni" antenna designs that can pick up every phone and other radio devices in the area (interference). These were fixxed by using directional antennas (but now, not a healing mesh). The other problem is self interference (when one mesh radio talks to another, but other mesh radios can also pickup the link). You can benefit by using 4.9 GHz (50 MHz band) to some degree, but this does not equal the capacity to do a very large scale deployment.

I have a customer who has a custom designed Fluidmesh system to connect 4 poles in a parking lot. Base radio on roof top with directional antenna to a "node". Directional antennas from there to the other 3 poles. All router radios, with antennas, are about $1500 each. I assume the base station radio is quite a bit more (with Skypilot it was $3500 base station, $1800 router radio, $550 client radio). This was with the minimal "bw" license. Directional antennas?? Really?? This does not seem to be self healing, which is how mesh is sold. $10,000+ worth of radios, when I figured I could do this with maybe $1000 PTP Ubiquit (nanostation, airgrid) or less with PTMP with better bandwidth and lower latency.

I am going to assume that networks (Firetide-Dallas??) ran into trouble when trying to use HD fixed and PTZ cameras (PTZ cameras without realtime control are frustrating and worthless). H.264 streams are efficienct, but they need a clean unrestricted path to work. I have found that the best way to use the SkyPilot 4.9 network nowadays is onboard SD storage (HD) and very low resolution RTSP streams back to central recording/management servers. PTZ preset control only. (As far as reliability, the (2) Gateway (12) Extender Skypilot system, with most radios at 60-120' on towers, has been working since 12/06. I think I have replaced one radio in 8 years due to a lightning strike - Amazing).

Dear IPVM Readers,

this is a very interesting topic and there is quite a bit of confusion out there on what mesh can and cannot do for video so I am going to share my thoughts on this matter hoping that it can help some of you in better understanding how to use wireless.

First of all, my name is Cosimo Malesci and I am Fluidmesh Co-Founder and VP Sales. We are one of the mesh companies mentioned in some of the above comments (which I found very informative and to the point).

First of all there are two types of mesh networks: the omni-directional mesh, and the structured mesh. The omni-mesh uses omni-directional antennas and every 'node' can talk to all the other nodes in range. This design offers maximum reliability but most of the time is not suitable for HD video because of the limited bandwidth (all the radios have to operate on the same channel and you are cutting bandwidth in half for every hop you do). In addition, omni-directional antennas have fairly limited gains so usually not suitable for most of the IP Video wireless jobs out there. The only exception here is mobility networks where we deliver connectivity to a fleet of vehicles such as an open pit mine, a port or terminal or a yard.

On the other hand, structured mesh uses directional antennas and it is a viable solution for video because it allows you to create a system that is still twice as reliable compared to a PTMP wireless network without loosing any bandwidth per hop. Each hop is a dedicated link on a separate channel so you don't have overlap. This design is the wireless equivalent of a fiber loop and has interesting applications in the critical infrastructure space where a higher level of reliability is needed. In addition, it is also very useful in situations where you don't have line of sight from every client to the main AP so you can create a linear mesh and branch out along streets or corridors with no need to run wires.

Of all the wireless systems we do, probably less than 30% of them use mesh in any shape of form. Most of the system we help our customers deploy are PTP or PTMP. These architectures most of the time offer a more cost effective solution and still plenty of reliability for customers. We run PTMP using a TDMA based protocol thus solving the issue of creating hidden terminals in the network.

Finally, I would like to clarify the confusion regarding the prices of our radios. Our radios currently start at around $500 MSRP each. Although we are more expensive than some of the low cost solutions out there, we use a much higher quality hardware, an MPLS Protocol which is not 802.11, 2 year warranties, and unlimited tech support by phone and emails. Although saving on the hardware cost is very attractive, there are some implementation and maintenance issues that I feel need to be considered before deciding what vendor to go with.

Hope this helps!

Cosimo

I would have to assume that using Fluidmesh, if cost was no option, would probably be a good solution. From pricing I got, it looked like $500 for end point radios and $1200 or more for multi-radio nodes plus antenna costs.

I have one site in a city with multiple PTZ sites (1.3 MP@ 15 fps each). I run MIMO PTP integrated antenna radios using the POE passthrough with careful selection of frequencies. On one chain, which has been running for almost a year, there are 4 PTZ cameras with 4 links. I use the POE passthrough eliminating switches at each site, so you basically just need a PTZ and 2 radios (1 at the end of the line, but because of the low cost of the radios, I have (2) with the unused one pointing to a standby radio on the fiber network in case I need to come in from that direction upon problems or for in car access). 8 total radios on this chain at a cost of $600 total plus mounting hardware. I have not to this point seen any dropped frames and PTZ control has no latency (Exacq with PTZ manufacturer listed with their PTZ priority that came out a couple of versions ago). Full bandwidth throughout without any self interference or time related issues due to mesh topology.

If this was Fluidmesh (or Firetide, etc), I would assume I would need an endpoint radio plus 4 multi-frequency radios plus antennas plus license key to unlock additional bandwidth. ($6000-7000+). Great solution made for this application (if costs were no object), but with the one listed above, equal performance to this point with no frame/camera drops.

As another note, I was expecting these cameras and radios to suffer vandalism or even bullet holes (which I bet the warranty does not cover). This is in prostitute and drug alleys with occasional murders. Replacing a $75 radio is more easy to stomach than a $1200 radio. The solution paid for itself in the first couple of weeks when you look at the arrests that were made for major drug busts. Amazing that none of the sites has been vandalized to this point.

Cost for all hardware (PTZ camera, (2) radios, nema box, AC harness to pole, camera license, surge protection, POE+ power, hardware) was $1200-1500 per site. Because the cameras and radios are generally on corners below building heights, similar frequencies can be reused with a planned layout/frequency/azimuth design.

Other Note: I run radio links also to various parking garages (multiple cameras at each of 6 garages) and various remote sites. (All other cameras on this network are via fiber to individual buildings - about 25 of them with 5-80 cameras at each site with (2) of these using AirFiber for 5 mile high speed links which also provides site network connectivity). SkyPilot mesh is used for Exacq Edge on remote cameras (ISD) running low resolution streams to the system with 2 MP Edge storage to SDXC and wifi radios for local system access and service if needed. Just about every location in this large city in a river valley is accesible by the 4.9 mesh network.