Home
About Me
Cheap wifi voice network
 


I've been experimenting with wireless voice solutions for a while and I think I should document a bit of it here.

What do I mean by wireless voice?  Not cellular, but using 802.11 wireless (wifi) to carry voice from devices to your telephone system.

There are a lot of issues here, such as wifi was never designed to carry 'time sensitive' data such as voice.  Also, there's an issue of 'handover' from one wireless access point to another - the time it takes usually resulting in a dropped call, (well, that's been my experience anyway).

If you are willing to invest in an 'off the shelf' solution, there are plenty tp choose from, (Avaya springs to mind), but what if you are on a limited budget and could use a wireless voice network, like a shipping department or a warehouse where people are always moving about?

You might say "Just give people cell phones" - but where I'm located even a basic cell phone seems to run about $35 a month, and if you give a phone to 4 people that's $140 per month, plus call charges (if any), plus relevant taxes.

About a year or so ago, I found out about a project called SMESH from John Hopkins University, (for the record I have no connection with SMESH what-so-ever other then the fact I think it's awesome).

Smesh is a 'MESH' based wireless network, meaning Access Points work together to form a 'Seamless MESH' of wireless coverage.  It's based on the OpenWRT operating system and is designed to maximize the speed at which data travels across the mesh and also minimize the amount of time it takes to handover devices from one access point to another.  It was designed with voice traffic in mind.

The part that really interested me was they had modified the kernal of OpenWRT to *really* speed up how data was handled and passed between access points (called nodes).

This enabled SMESH to be installed on cheap Linksys WRT54G/GL access points. The SMESH guys even supply a modified WRT firmware to make life easier.

I will admit, I was not sure it would work but it was cheap to test.  I purchased 3 WRT54GL units (approx $200), installed the modified firmware and after some learning and playing about, I got the mesh up and running.

My 3 access points were arranged in one long line, (1-2-3).  Where the signal strength of an access point fell to 50%, I installed another access point.  It was highly unscientific...   :-)

When I was stood at access point 3, I could be certain that I was not connected to access point 1 and visa versa.

In the config of SMESH, you define which access point will be the ‘gateway’ to the Internet or your LAN.  I connected this gateway to an Asterisk server.  Now I could use a wireless client to connect a SIP client to Asterisk for testing.

I used a Windows Mobile cell phone (Version 6), enabled the built-in SIP client, (Windows Mobile phones do have a built-in SIP client but you have to jump through some hoops to get it working), registered with Asterisk and placed some calls.

I established a SIP call and walked from access point 1 to access point 3.  No problems, call continued without interuptions or dropping.  I turned off access point 3 and repeated the test.  Just over halfway from access point 2 to access point 3, the call dropped - proving that the call before must have been using access point 3 to continue.

I've now been running these access points with SMESH for about 6 months.  There are 4 parts guys using Nokia E63's cell phones, (more on them in a moment), moving about a shipping department and a warehouse - oh, did I mention the warehouse is in a separate building about 250 feet away...  :-)

The guys wonder all around and between buildings with no dropped calls.

Thanks to Asterisk being connected to the existing Panasonic phone system, the guys can be contacted directly from the outside work on a regular phone number and can make internal or external calls.

Yes - it works fine with OCS and Lync, (assuming the IP telephone system your mobile SIP clients connect to is also connected to OCS / Lync).


But what about outside wireless coverage?  You can buy waterproof enclosures for WRT54 units and remove the original plastic case - but these usually cost more then the WRT54 unit itself.

Me?  I purchased a large plastic container from Walmart for $4, drilled a small hole in it for power, sealed the hole when the cable and the WRT54 was installed and glued the lid shut.  It has now survived six months, a minus 15 C winter and Vancouver rain - I'd say it's doing OK.  ;-)


So - it's all fantastic and you should rush out and do this right now?  Well no, there's some 'gotchas' and here they are...

- SMESH works using wireless 'adhoc' or 'computer to computer' mode.  This means cetain phones will not work with it, for example, Blackberrys will not work on an adhoc wireless network and neither will the Polycom range of wifi phones (shame - because those Polycom wifi phones are very good).

- Android phones will work with adhoc mode (with a hack), Windows mobile phones and Nokia phones, (Warning: only certain Nokia cell phones have the built in SIP client - check here which ones have the SIP client http://wiki.forum.nokia.com/index.php/VoIP_support_in_Nokia_devices)

- Of the phones I've tested, the Nokia phones are by far the most user friendly for the average user.  I buy E63's for $75 each on Criagslist, plus $80 for a Mugen 3600mAh battery.

- Using the phones like this seems to *DRAIN* power, buy an extended battery.

- SMESH has no QoS, all traffic is treated equal.  I'd use your SMESH network for voice only and use normal data on something else.


If you can live with the 'gotchas', SMESH does provide a wireless voice infrastruture for dirt cheap.

I've been asked buy Nokia E63's as fast as I can, expand the install to cover the entire complex and to implement it at other locations...


There's also another wireless network I've recently become aware of called WING.  It requires more powerful hardware, but promises to work using a regular wireless infrastructure, not adhoc, and can support multiple radios, (one for the mesh and one for back hauling data), and offers QoS so you can use it for regular data as well as voice.  It seems to be about $200 to $300 per access point.

There are several people using it saying it works very well, but it does require a little 'tuning'.  I plan to experiment with it when I can get some time and free up some money for hardware to test with.


Hoped the rambling helped...

Please feel free to contact me using paul [at] pauladamsit [dot] com


  
Top