Greetings,

We are a group of Ham operators in Malta who have been experimenting with HAM BROADBAND for a few months now and found it to be absolutely amazing. There is a learning curve, as with everything, but so far so good.

We have set up network tests using FTP, Web Services, Streaming Services (HD cams) and in the future plan to implement VOIP for emergency amateur use. All worked fine for us so far, but did find an issue with speed.

Prior to continue building the MESH, we are testing SPEED limitations and we have found out that we we cannot transfer anything (in controlled ideal conditions) faster than 15Mbt/s (that is: at a 1.5MB/s approx). We have tried the FTP transfer experiment between two nodes and yet, this is the maximum speed we're achieving.

We are using LINKYS WRT54GS as well as UBIQUITI M2 (also 2.4GHz). The Ubiquiti's performed at the same rate of 15Mbt/s, just like the Linksys, when we know that the Ubiquiti devices can perform at a MUCH faster rate (on their commercial original firmware). The Uniquiti devices have a directional 'dish type' antenna and they are POE's with 24V, whist the Linksys were tested using their original rubber ducky and voltage supply.

Hence our queery ... what is throttling us? Is it the Ham Broadband firmware? Is the implementation of 802.11 used in Ham Broadband? Can't we go faster than 15Mbt/s (1.5MB/s FTP transfer rate) on 2.4 GHz? Do we necessarily need 5GHz equipment to achieve faster and better bandwidth than this?

We appreciate any help and/or opinions on the matter. All our nodes are set up in the mode: MESH NODE.

Indeed.

Which is why we also tested UBIQUITI to UBIQUITI alone, without any linksys on the mesh and yet we achieved the same speeds achieved when we tested the mesh using only linksys devices.

This cannot be because we know ubiquiti to ubiquiti transfer rate should be much faster than linksys to linksys.

We cannot understand why uniquiti is still performing at 15Mbt/s, just like when we tested linksys to linksys.

As noted, Stock AirOS will go up to 40mhz wide channels that will be part of it as we are not currently supporting 40MHz Channels (the hardware will do it but we haven't seen much need for 40MHz yet, if anything it's the 5/10 that have the focus right now)

MCS Data rate should kick in however also note that other wifi traffic can slow it down.

I haven't had time to do a benchmark lately so I'm not sure if you are seeing limitations in your setup or some other item.

It's a complex issue.  Here's an article that could shed some light on mesh throughput and the rates you've been seeing: http://centmesh.csc.ncsu.edu/course_report2.pdf

Andre, K6AH

Complex it is indeed.

Thanks for your continuous support on the fantastic Ham Mesh and the assistance given through these forums. It is indeed a pleasure to see it work.

The intention locally is to set up ham broadband networks for emergency use, with VOIP and HD cams operating at the same time, hence our hope for more bandwidth, not at the cost of network instability either. I guess we are all pioneers in respect to experimenting with the system.

We know that the mesh can be fortified and made more fault-tolerant with every node added to the mesh itself. The problem is that we don't have much nodes to play with for now, so most experiments were mostly point-to-point in essence.

I don't know then, if hypothetically speaking, if we had 40 MHz bandwidth instead of 20MHz, we could get better distances at the same maximum 15Mbt/s we managed to achieve so far....

I'll need to read into the document shared with us by K6AH. It's certainly a great read into how mesh behaviour changes, sometimes quite in a weird way, depending on network topology and node placements and a ton of other factors.

WIFI noise is one that we have noticed for sure, but going to a more remote place where we couldn't detect any commercial WIFI anywhere, the maximum throughput in ideal conditions was still 15Mbt/s  - which is presumingly caused then, by the 20MHz bandwidth limit.

I've read the document that k6ah shared with us. It is certainly a fascinating read showing various types of throughput behaviour on meshed networks.

However, as stated in the document, the experiments performed did not consider OLSR, which is the routing protocol that is maximising throughput on ham broadband. Also, since 2009, it is safe to presume that several enhancements have been made since then to the wireless equipment itself.

But as said, it is an excellent read and experimentation done by North Carolina state university.

For multi user VOIP and hd cams to work continously in an emergency setup, we need more data bandwidth.... or more and more nodes. Is our logic correct?

We also experimented with different polarisations and directionality. So far, the results fly in the face of traditional radio where it seems that we are working faster with omni-directoonal than with directional antennas.