Hi all,

I'm researching on this topic for some time now and I'm interested to see what you guys think about this topic and during time, I hope, this could be a good tutorial for all the others who stumble upon this problem. If you dont have any experience, you can also contribute by posting some relevant links or by discussing about this in order to clear up most of things regarding this issue. After we finish discussion on this topic, we can sum up all the facts into a FAQ that will provide others a quick guide for future use.

So, let's start. The goal is to connect 2 geographical points with a link that has a good scalability, meaning, we can add bandwidth easily, when needed.
The first thing that comes to mind is to setup several parallel links, each one using a different radio channel. But, in order to reduce the need for new antennas, we can try to combine all those radio links and transmit them over one pair of antennas. Ideally we could achieve N links over 1 communication channel, without any loss, thus avoiding expenses of mounting new antennas on windy towers, whenever we want to add a new link.

I've realized by now there are several devices that can be used in this scenario as combiners of radio links (like diplexers, triplexers, multiplexers, combiners, splitters, etc.). Their purpose is to utilyze a single pair of antennas to transcieve several logical radio links (using different radio frequencies). As I research further more, I also realize that so many times announced 802.11n standard supports MIMO (Multi Input Multi Output), which basically uses antenna arrays to achieve many-to-many links (correct me if I'm wrong).

Another approach to this is using a dual polarity dishes (antennas). Shortly, these kind of antennas have 2 heads that can transcieve 2 radio links over 1 pair of those antennas, without the need to use any of those combiners. However, our goal is to establish several links (as much as we can) to utilize 1 physical pair of antennas, so, even suggestions that involve using dual polarity antennas are welcome in this discussion.

I've also just found out that, recently, guys from SkyCross have developed an 3 feed antenna, which allows 3 input signals to feed the single antenna. That's what they call IMAT (isolated mode antenna technology). More info here: http://www.skycross.com/Technology/iMAT.asp

So, if you have any thoughts on this topic, please, feel free to discuss it here. Also, any relevant links to other topics on this forum, device manufacturers websites, tutorials on the internet and other relevant stuff is highly appreciated, too.

Here are some of the links I've found on this forum:
- NSTREME 2 on Single Rootenna?
- diplexer
- Diplexer, Multiplexer and power splitters...
- Can frequency diversity be made with mktik?
- Connecting two radios to one antenna
- diversity Rx Tx selection
- Design Problem

Also, here are really worthy posts I've found googling around:
- multiple radios one antenna
- Multiple WiFi Channels on one Antenna

Thanks to all who participate in this discussion.

Keywords for this topic (for users searching this topic):
duplexer, duplexers, diplexer, diplexers, triplexer, triplexers, multiplexer, multiplexers, combiner, combiners, splitter, splitters, antena, one antenna, single antenna, two antennas, more antennas, several antennas, one link, single link, two links, several links, more links, parallel links, frequency, frequencies, frequency separation, channel, channels, chanel separation, filter, filters

The "combiner" costs more than two additional antennas + time to mount them.

Use a dual polarized antenna for having two links on one antenna. That's the best you can do at a reasonable cost.

Sometime it's just not possible to mount several antennas at one geographic point (for ex. when renting antenna tower's space), because you don't have space for new antennas or you are just not allowed to put several antennas, etc.
I believe that 802.11n standard has approached to this idea and I'm glad it did, because it is using the spectrum more efficiently now, but knowing that MicroTik guys have failed to implement the support for that standarad for so long time so far, I really doubt that waiting for them will be a good idea. I guess something really bad happened with their project and I'm afraid it's not gonna happen any time soon (they were announcing it a year ago, and still nothing?)..

Also, I've been reading about some additional devices that can combine those signals, prior to sending them to the antenna and as soon as I sort things out about those, I'll post the conclusions here. One thing I need to mention, of course, it's obvious that the most crucial part of the problem here is the transmitting part of the antenna. That means our main goal is how to combine multiple channels over one physical medium of transportation (antenna). And I believe that receiving multiple channels/signals on one antenna is not something complicated, because each wi-fi card, connected to that antenna, filters out things out of its selected channel (frequency), so, even if antenna receives multiple signals modulated on multiple channels, each card will filter out things it doesn't need and leave only the desired channel/signal received.

Very interesting, however putting more then one signal on an antenna from multiple radios will in effect create enormous amounts of 'noise' for the other radios. The 802.11 radios do not filter out of band iterferance very well (try setting up a pair of links and set the frequencies 20 Mhz apart, then run speed tests on the links individually and simultaneously). You would most certainly have to use bandpass filters after the radio and prior to the combiner. This would add even more cost to the setup, and while I believe it would work, this type of thing would really only be needed if space was at an absolute premium, and you really had to use 802.11. Alternatively, with all the extra costs you could buy a 'premium' radio system capable of higher speeds across the single antenna thus using the spectrum more efficiently.

Still an interesting idea.

Cheers

Seems like you are solving the wrong problem. Dont worry about combining RF, it wont work because of previous comments regarding noise.

If you need to get 3 independent streams of data between 3 pairs of radios, solve it with RouterOS.

Connect the two 'localities', if this is the scenario, with a backhaul such as 2 R5H based radios on a 411 for example.

Of course you must collect the 3 streams from each 'locality' by using an AP to allow them to connect to the backhaul location.

Final step. Create end-to-end tunnel of your choice. Use tunnelling as the 'combiner/duplexer'...

Since the 'combined' RF would take the same path as the 'tunneled' streams, addition of two APs should solve the problem unless geography or topology precludes connection of all nodes on each end.

Just to remind, "The goal is to connect 2 geographical points with a link that has a good scalability, meaning, we can add bandwidth easily, when needed". So, znet, if I understood your post correctly, you proposed a way of just combining several links through one way, but this way you don't get 3 times the max bandwidth of each link, instead you are sharing 1 max bandwidth (depending on frequency used, the distance of the link, etc.) on those 3 links, right?

Also, I agree about the noise, since that's the main problem with dual pol. antennas too, so I've figured it's something that cannot be easily handled, but it's not impossible. Anyway, if I'm not wrong, if we ever achieve this goal in ideal conditions (without a significant loss/crosstalk/etc), we would still be able to use only a few of available channels, over one physical link.

Consider the next image:

The image is showing the channels for 2.4 GHz spectrum.

From the image we can see there are max 4 channels that can be used at the same time without interferrence (channels 1, 5, 9 and 13), so in an ideal situation, we could have max 4 links combined together on 2.4 GHz spectrum. Now, these channels are all 22 MHz wide, so there would be some overlapping, but we will ignore this, just for the sake of simplicity. So in our case of 4 combined signals, we would have the same effect as if we could have 1 signal, centered on channel 7, with the width of 80 MHz (the whole 2.4 GHz spectrum). After all that's how turbo mode works, right? It expands the channel width to achieve more bandwidth.

So, my question is this, is there any chance that we could achieve our goal with a simple channel widening, rather than by having 4 links combined with some complicated equipment? Are there any significant negative consequences when using a wider channel (40+ MHz) in comparing to a channel 22 MHz wide? Are there any legal issues or such?

I ask this because I've stumbled upon some devices that use wider channels for ptp links, operating on 2.4GHz/5GHz, but each of those use their own proprietary protocol like "proprietary frame-bursting, compression, channel bonding and point-to-point side session technology to improve IEEE 802.11g wireless LAN performance". Please see this table at Wikipedia for more info: http://en.wikipedia.org/wiki/List_of_de ... s_networks and pay attention to the item named "802.11g with Nitro (Conexant-proprietary extension to 802.11g)". It claims it's capable of 140 Mbit/s throughput transmission speed. Did anyone have any experience with such devices?

I mean this has to be possible, because Motorola did something like this already. Take a look at this page: http://www.motorola.com/ptp. There is a model called "PTP 600 Series" that claims this "Point-to-Point wireless Ethernet bridges operating in the 2.5, 4.5, 4.8, 4.9, 5.4, 5.8 and 5.9 GHz bands at speeds up to 300 Mbps." how do they do that?

One last thing. I was hoping 802.11n will help with this issue, but I've read that MIMO is not good for longer links, it's only usable on shorter distances (like a few km or so), so I guess 802.11n will not help us much with ptp links over 10km. Please correct me if I'm wrong.

Just to remind, "The goal is to connect 2 geographical points with a link that has a good scalability, meaning, we can add bandwidth easily, when needed". So, znet, if I understood your post correctly, you proposed a way of just combining several links through one way, but this way you don't get 3 times the max bandwidth of each link, instead you are sharing 1 max bandwidth (depending on frequency used, the distance of the link, etc.) on those 3 links, right?

Also, I agree about the noise, since that's the main problem with dual pol. antennas too, so I've figured it's something that cannot be easily handled, but it's not impossible. Anyway, if I'm not wrong, if we ever achieve this goal in ideal conditions (without a significant loss/crosstalk/etc), we would still be able to use only a few of available channels, over one physical link.

Consider the next image:

The image is showing the channels for 2.4 GHz spectrum.

From the image we can see there are max 4 channels that can be used at the same time without interferrence (channels 1, 5, 9 and 13), so in an ideal situation, we could have max 4 links combined together on 2.4 GHz spectrum. Now, these channels are all 22 MHz wide, so there would be some overlapping, but we will ignore this, just for the sake of simplicity. So in our case of 4 combined signals, we would have the same effect as if we could have 1 signal, centered on channel 7, with the width of 80 MHz (the whole 2.4 GHz spectrum). After all that's how turbo mode works, right? It expands the channel width to achieve more bandwidth.

So, my question is this, is there any chance that we could achieve our goal with a simple channel widening, rather than by having 4 links combined with some complicated equipment? Are there any significant negative consequences when using a wider channel (40+ MHz) in comparing to a channel 22 MHz wide? Are there any legal issues or such?

I ask this because I've stumbled upon some devices that use wider channels for ptp links, operating on 2.4GHz/5GHz, but each of those use their own proprietary protocol like "proprietary frame-bursting, compression, channel bonding and point-to-point side session technology to improve IEEE 802.11g wireless LAN performance". Please see this table at Wikipedia for more info: http://en.wikipedia.org/wiki/List_of_de ... s_networks and pay attention to the item named "802.11g with Nitro (Conexant-proprietary extension to 802.11g)". It claims it's capable of 140 Mbit/s throughput transmission speed. Did anyone have any experience with such devices?

I mean this has to be possible, because Motorola did something like this already. Take a look at this page: http://www.motorola.com/ptp. There is a model called "PTP 600 Series" that claims this "Point-to-Point wireless Ethernet bridges operating in the 2.5, 4.5, 4.8, 4.9, 5.4, 5.8 and 5.9 GHz bands at speeds up to 300 Mbps." how do they do that?

They do mimo and use a software defined radio (not Atheros based). They use only 30MHz for
up to 300MBit real traffic. You can buy a device for 5.4 or 5.8 and then you are fixed to this band.
Dont expect to buy a cheap miniPCI card and achive the same results.
And dont compare these radios. Its a complete different price-range. And you need MT in addition
as they are simple bridges.

One last thing. I was hoping 802.11n will help with this issue, but I've read that MIMO is not good for longer links, it's only usable on shorter distances (like a few km or so), so I guess 802.11n will not help us much with ptp links over 10km. Please correct me if I'm wrong.

Stefan

Regarding higher than normal bitrates, the Moto solution, and any other solution that claims higher bandwidth, they have one common characteristic and parameter that they use to accomplish their high bitrate task: ITS THE MODULATION!

According to the 802.11a spec, the highest order modulaton is 64QAM. Didnt take vendors to realize if you use 256QAM you would get 4 times the bandwidth. That company with the red line of equipment takes the highest base rate to 72Mbps, and then can 'turbo' it to 144Mbps raw bitrate, but dont know off hand what they did. However! The high symbol rate of the 256QAM based carrier (per se) isnt as tolerant of rough conditions. Accordingly, much licensed equipment has been doing this for a long time to produce inconceivable bitrates. The very narrow antenna beamwidths, coupled with kilowat link budgets (~60dBm EIRP) put the signals at the maximum a receiver front end can handle, therefore yielding insignificant BER as a factor of the sky high SNR.

So, the moral of the story really is, you get what you pay for. The concept of multiplexing microwave is analogous to DWDM in the fiber world. They keep stackin unique colors to linearly multiply the bandwidth. Those cell carriers use the metric 'bits per second per hertz' to measure the efficiency of different transmission schemes. So I will let someone else figure out that metric for the proposed multiplexed solution. Then the config of the RF 'plexers' can be determined, and more importanty deermine if it physically can be done. If it can, then figure out the $ per Mbps. That could make a workable solution totally useless because of cost.

Thats my story, and I am stickin to it...........