Hi all, I'm looking for some good recommendations on setting up an
802.11b wlan to coover a warehouse that is roughly 120,000 sq ft in size. The ceilings are roughly 30ft high and the warehouse is filled with floor to ceilng racks holding boxed and palletized product. The purpose of this WLAN is to allow the use of handheld 802.11b scanning equipment to scan incoming and outgoing shipments into the warehouse management app. We are currently using wired scanners, which are located in one area at one end of the warehouse. This makes doing a full inventory or rescanning pallets very difficult.
Currently I have looked at the 3com MAP's and Wlan switch. These units seem to cover all of our needs, as they do built in PoE and allow centralized configuration. The downside is the cost. While I don't expect this project to be done for free, $2600 for the switch and 300+ each per MAP (estimating roughly 8 MAPS needed) might be a hard sell to the powers that be.
I've also noticed that D-link makes MAPS and Wlan Switches that also do PoE. But I'm weary about going with Dlink for something that needs high availability.
I am aware that I could also go with several standard AP's and configure them with the same SSID's and Key's and the devices could just roam between them. I have done this in the past in situations where mutlipe AP's were needed for coverage reasons and I had no significant problems. The downside to this is the individual configuration and lack of integrated PoE for most devices of this nature. The upside is, naturally, the cost.
Can anyone make some good suggestions on either products or methodologies to cover what I'm trying to do here? I already have some good ideas that would accomplish the goal. I"m just looking for suggestions or ideas from those who may have already ventured down this road and might have more direct experience.
140,000 is only about 380 foot on a side. So you need a range of about half that. Humm. Eight units sounds like a lot but I'm no expert. All that racking might cause a lot of problems. Might be worth experimenting with a consumer grade wireless access points and a decent aerial. Borrow one and stick it in the middle as high as possible and see what the signal strength is really like in practice. Does it have to be 802.11b? I think g has better range but maybe mistaken.
It depends on how well the 2.4GHz signals penetrate the palletized/boxed product. With good signal penetration, a few stationary WAPs might be able to cover all the aisles. With poor penetration (which is rather likely), you may want to consider a whole different strategy of buying portable (handheld) inventory recorders instead of an online system.
This is also known as the ultimate maximum reflection environment. I've helped setup wireless in a similar (cold storage) environment. Conventional wisdom is to "illuminate" the aisles. We used Cisco 340
802.11b access points, which are rather crude by todays standards.
The big problem turned out to be mutual interference between access points. We initially installed them high, near the roof, in order to cover the greatest area. All they did was interfere with each other. When they were moved to about 10 ft off the floor, we lost coverage, but the interference was less.
We also had a big problem with seamless roaming. The protocols simply don't exist to do it right. Each company has a proprietary solution. The inability to move automatically to the "best" access point was a serious problem covering lunatics that would ride the forklifts down the aisles taking inventory with a wireless bar code gun. It also interfered with the bar code gun "laser tag" games. Since the seamless roaming requires cooperation in the client radio, using a mixed mess of clients was not a great idea (which we did anyway).
Good. The scanners are not moving. That avoids the roaming problem.
Again, if you are moving, you're not going far. One company ended up with a computer on a roll around cart. It had an 802.11b wireless connected computer and wired bar code reader attached. The cart was always buried in paper. Not as fast as riding the fork lift, but more reliable.
Don't use repeaters. They only cause interference.
Check that your software doesn't require a 100% reliable connection. The stuff I was working with would hang if even one packet was lost.
Verify data reception. The operators would pull the trigger to scan and nothing would happen. So, thinking they missed the bar coded target, they would do it again and again and again. When they moved back into range of the radios, the reader would empty its buffer and record the same entry as many times as they pressed the trigger.
Spend some effort getting proper antennas. Most wireless switch vendors do not provide external antennas and have an omni pattern. That's usually fine, but not for illuminating a narrow aisle. You need a directional antenna, that will concentrate *ALL* the RF down the aisle and nowhere else. 8dBi to 14dBi panels seem to work with
6ft aisles. IMHO, the antenna is the most important part of the puzzle.
I cringe every time someone mumbles "self-configuring" and "self-optimizing". That's a guaranteed service call for "nobody changed anything but nothing works" type of call.
All the wireless switches are expensive because they're all proprietary. I think a wireless switch with PoE is the way to go. It solves the number one time burner problem, central system management. It's something nobody thinks about until the problems start. You either pay for centralized systems management at the front end, or burn the IT labor fixing the mess later.
Look into other wireless switch vendors offerings: Symbol Aruba Cisco Entrasys Xirrus Foundry Networks (probably more vendors) I haven't been keeping up to date on Wireless Switches so can't advise on specific products. There was quite a bit of new goodies product released at Interop last week.
I dunno about this one:
$2500, it looks much cheaper than the others. No experience with it. I don't like the looks of the circular access points. 802.11a/b/g is nice. No obvious external antenna connection. "Zero-Config Installation"??? I'm worried.
I've had problems with DLink products in the past. In general, if it doesn't work out of the box, it's not going to get fixed. That's the real problem. All the vendors have their problems, but only a few bother to fix them.
Yep, that's about it. All your eggs in one vendors basket, or a management nightmare. There's no in between. Toss a coin (or toss your bank book) to decide.
Well, I've only helped on 2 warehouses. My bag is RF and antennas. Layout, location, and antennas are serious problems in a highly reflective environment. My contribution was mostly to move the radios down and add proper antennas. I also reduced the number of access points in the building from about 14 to 8 to reduce mutual interference. Beyond that, the applications people fixed their program so that it could tolerate connection outages. We never did fix the roaming problem. Also, I supplied the giant stickers with the front and back side bar codes for the laser tag games.
Have you considered using an AP with a distributed antenna? Run the antenna along the ceiling in a manner that best illuminates the aisles. I've never heard of distrubuted antennas being used for WiFi, but it shouldn't be a big deal to make one if they're not commercially available at this frequency. It's basically a leaky coax like the distributed antennas used in tunnels to serve vehicles passing through.
I've used power splitters and they do work in very specific applications. You run a power divider to the access point and connect one antenna on each port. I had one system running a 4 port power divider. One directional antenna on each aisle should illuminate the required area.
However, there's are problems. The power divider reduces the signal by half for each 2 ports (plus about 0.5dB loss per port). A 4 port power divider has about -6.5dB loss per port in transmit, and about
0.5dB loss in receive. That's less than half the range. With a reasonable amount of TX power to compensate for the loss, it should work.
Another catch is that the antennas have to be arranged so that they don't see each other. That's easy enough to do on a radio tower. Just point the antennas in different directions. However, in a closed warehouse, the isolation is going to be minimal and the reflections are going to be horrible. There will be nulls and inter-symbol interference, both of which are fatal. It can be made to work with careful antenna selection and positioning, but I'm not optimistic. The one indoor system where I used a power splitter had to revert to individual AP's, on different channels, in order to get things to work well.
Generally, trying to cover the same area using multiple antennas off a splitter will not work reliably because of position-dependent phase cancellation as you mentioned. In fact, one technique commonly used to build antenna arrays with shaped patterns is to drive multiple antenna elements off a splitter. Their patterns are due to controlled signal cancellation.
I think he's referring to Andrews Radiax or Cablewave Radiaflex.
140K sq ft - that's 374 foot square, or 350 x 400, and so on. I would think the coupling loss would be a bit to high, but _otherwise_ this might be worth have a manufacturer's application engineer talk about it. It's not something you want to 'buy and try' - the other way round might be better, but I don't know how many manufacturer's reps would take the gamble.
OK. I can do that. I don't think it will work. Let's grind the numbers.
It will be a long time before I find a manufacturers rep that can do calculations much beyond the lunch bills and expense reports.
PREDICTION OF INDOOR WIRELESS COVERAGE BY LEAKY COAXIAL CABLE USING RAY TRACING
Lunch bills and expense reports are easier. Never mind.
So, let's do this one with Andrew Radiax. The smaller and cheaper RCT "radiating mode" coax isn't rated at 2.4Ghz. You have to go to 1 1/4" coax before it can be used. So much for RCT coax.
1/2" RLX RXP4-1 coupled mode should work better:
$4.10/ft from Tessco or $6,500 for just the 1600ft of coax. Seems a bit overpriced, but that's what the distributor wants. Are you *sure* you wanna use leaky coax?
Example of how to calculate a cellular link budget with Radiax.
Basic specs at 2.4Ghz for RLXP4-1 is 6.7dB/100ft and -62dB coupling loss at 6.5ft. Let's do 2 access points with a 2 way power divider at midpoint using 4ea 400ft coax segments. Diving in:
TX power: +20 dBm (assuming a 100mw radio) Coax feed loss: -3 dB (including connectors) Power divider: -3.5dB Cable loss: -26.8dB (400ft at 6.7dB/100ft) Coupling loss: -62 dB (at 6.5ft) ======= Signal level: -75.3dBm at 6.5ft.
Perverse square law says that power drops -6dB every time we double the distance. This thing has to work half way across the building which is 400ft across. At 200ft the range is: 200 / 6.5 = 30.8 times as far  20 log (30.8) = -30dB
So, the signal level in the middle of the warehouse is about -105 dBm. That isn't even close to functional. The average 802.11g laptop, at
11Mbits/sec connect speed, requires at least -65dBm to function. Add about 15-20dB for a decent fade margin. This leaky coax model is at least 55dB short of being functional. That's a HUGE amount.
What went wrong? The large coax loss through 400ft of Radiax certainly didn't help. It would probably barely work in close proximity to the Radiax, perhaps to 10ft, but not much more. Why does it work inside buildings and tunnels? Very few try leaky coax at 2.4GHz for extended distances. Most systems are at cellular
800/1900MHz frequencies. In the offices, they go in suspended ceilings, which are fairly close to the users. In tunnels, the coax is on the wall of the tunnel, which is also fairly close to the train. In buildings, the coax is run above the hallways, again close to the users.
One could do the same with the warehouse except that the ceiling is
20ft high, not the usual 8ft suspended ceiling. That's not as bad as
200ft away across the warehouse floor, but still far enough to reduce the signal level to unusable.
Incidentally, I've re-discovered that mounting Radiax against a concrete wall is a bad idea. It's really a long antenna and doesn't like objects in the near field. We had to use spacers to keep the coax away from the walls and pipes. Also, sharp bends tend to tear apart the coax at the slots.
The warehouses that I cleaned up had this problem. That's where I found that 802.11g is far more tolerant to dealing with reflections, multipath, deep fades, nulls, and other cancellation/reinforcement phenomenon. When I fixed the wireless connect speed at the access points to 12Mbits/sec, the fade/null problems were greatly reduced. However, the ultimate solution was to position and specify the antennas so that they did NOT see each other, thus preventing these effects.
 I have this handy table on my wall mostly for inverse square law guesses. Double the distance and lose 6dB: dB = 20 log (multiplier)
Distance Power Loss Multiplier dB 0 0 2 -6 4 -12 8 -18 16 -24 32 -30 64 -36 128 -42 256 -48 512 -54
Rebar, unistrut, conduit, CAT5 cables, telco cables, emergency lighting, metal shelving, water pipe, drain pipe, vent pipe, air manifold, and the usual mess of bulletin boards, calendars, wall hangings, and Playboy calendars. I used a TDR (time domain reflectometer) to see if there were any "lumps" along the line. If I can see it with the TDR, it will be truely horrible at 2.4Ghz. Empirically, I needed about a wavelength of spacing from any metal object to have it not show on the TDR. It was really amazing how I could see every piece of metal that crossed the leaky coax. I also tried covering any crossovers with about 6" of aluminium foil duct tape. That actually worked quite well and really improved the TDR picture, but it was messy, difficult to install, wrecked the waterproofing, and looked really sloppy. Spacers were the best solution.
Incidentally, my previous calculations are the worst case radiation from the end of a 400ft Radiax run. The signal level will be considerably higher it were calculated closer to the access point. Unfortunately, my usual luck demands that the most important client radio will be located near the end of the coax, and therfore in the worst possible RF location.
Aww, come on! There (at least) used to be a rep up in Mountain View... found the coffee cup - Allis Associates - on Castro between Villa and Dana (block South of the rail station) who did RF stuff and did some technical advising. They weren't _that_ bad. (They used to love me - the NASA CTR was cheap, and wouldn't let me buy in quantity, resulting in lots of petty-cash purchases from the reps.)
especially if you have a computer with a fancy spread sheet app.
Heck no! I know better. I was just steering you for the O/P.
I don't have my Andrew or Times books any more, but yeah that sounds about right.
That's the killer
7 dB over a wide-band grunt.
Yeah, that dead before the RFQ.
I've seen it used at other bands, but they have always been at relatively short distances from the cable, or higher power/narrow band stuff.
Another solution for really difficult sites is circular polarization. The problem with that is that the antennas are FAR more expensive, and really not practical for hand-held gear. You still run into problems with double bounces, and pattern distortion making things more elliptical than circular.
I taught the "new math" mode, and hammered in memorizing just 3 numbers:
3, 5, and 10 dB (double, sqrt 10, 10 respectively on 10 log). 4 dB? Simple, up 10 dB (10 x), down 3 dB (half), down 3 more (half again). 8 dB? 5 + 3, and so on. Range, voltage or current, verses power? Double the dBs. For hand waving (and most calculations), it's accurate enough, and needs only limited training to master. There's an old TI-30 scientific calculator in the briefcase if I don't have a computer and the answers need to be more "accurate".
No. I can run Radiax across bulletin boards, employee policy drivel, wall hangings, potted plants, company announcements, pipe, conduit, plumbing, and even windows, and nobody will complain (much). However, run Radiax across the Playboy or Rigid Pipe calendar, and the wrath of the entire warehouse staff will be upon thee. Some things are considered sacred.
That's cheating. Anyway, most warehouses are rectangular with depressing boring flat parallel walls.
OK, I'll conceed that there are a few exceptions. I even know a few that can read their own data sheets. Most manufacturers reps will immediately defer anything that requires more than simple arithmetic to the factory applications engineer. I used to use that to my advantage. I made sure I asked a techy question that went over the reps head. He would immediately drop the name and phone number of the factory apps people, who I would later call. If I asked the rep for their names and contact info, he would invariable refuse to supply them.
I just fix computers. I don't use them.
Gee thanks. Remind me to bait my next technical pontification with obvious bad ideas and guaranteed failures. I don't really need more entertainment value.
I've used Radiax at 2.4GHz once with limited success. I've also made my own with heliax and a rotary grinder. I think I could have done better with properly configured antennas.
Yep, sorta. I've done circular polarization on both point to point links and with access points. It's really nice for point to point because any odd numbered reflections are cancelled. Incidentally, I have two large helical antennas stored that I really wanna try. The real advantage of circular polarization is that it dramatically reduces deep fades. This is particularly useful for moving clients. That bad news is that the polarization loss between circular and linear polarization is -3dB. Not a huge amount, but still important.
Incidentally, I did some field testing at 460Mhz in Smog Angeles many years ago using circular polarization at the repeater site. It worked really nice, but the guy the wrote the test report made it sound like magic. Nobody believed it and I couldn't sell anyone on switching antennas.
Nice way to do it. I learned the Trachtenberg speed system of arithmetic quite early. I could easily do simple math in my head. Then, I slammed into more complex functions and found that I had to resort back to conventional methods of doing the math. Later, I bought a calculator, which ruined my mental math abilities. These days, I grind everything through my HP41CX and still make order of magnitude screwups.
Ah, the good old days, when the calendars showed something more interesting than the day of the week. 'round here, management is so afraid of sexual harassment lawsuits that a lot of the decent calendars are banned. The one in my office this year shows pictures of lighthouses, while the guy across the hall has what looks to be an Arizona Highways one. Nice, but definitely lacking something. But then, when the policy came out several years ago, someone had what looked to be a giant smiley on his wall. Took about three months before someone topside realized it was a classic Xerox of a moon shot. There was hell to pay for that one. Even the Disneyland mouse ears were not considered appropriate.
The really good reps can even _spell_ 'clue', while the better ones merely have heard the word used in a sentence. In the 80's, I had a really great one in Cupertino who probably could have made a fortune as a conslutant. He'd answer questions about the product field, even if his product line wasn't suitable. I steered a lot of business his way just for that reason.
I wondered why you ignored that section of his post.
You mean like the radio setup Ames was using inside the original 40 x 80 wind tunnel? On the old section, it was asbestos cement corrugated sheeting (which is also what Hanger 1 is covered with), but the reflections were "interesting". Nice slow wave structure at 2 GHz
The 1970s era microwave landing system chose vertical polarization because they felt they could get away from reflections due to the time verses angle basis of the scanning beams. Still, they had a test requirement where they used 35 x 70 foot wire mesh reflectors mounted on a trailer to induce multipath. The thing had to work with the reflected signal just 3 dB down from direct. We did a demo in Atlantic City with an earlier system that used circular polarization. Couldn't even see the effect of the reflector. Next test was a LHCP antenna sticking out one end of the van, 46 dB gain TWTA, variable attenuator, and a RHCP antenna sticking out the other end. Things still worked with the multipath (measured in the target area using a linear antenna) 22 dB greater than the direct signal. Didn't get adopted due to the higher cost of the antennas.
You get that loss with two linears differing by 45 degrees - worse when the klown with the Motorola HandiTalky is holding it horizontally when he talks into it - cause it looks kewl.
That happens. There was a similar bit with TVs using elliptical to reduce the ghosting. I know some transmitters got it working, but no one ever bought the appropriate receiving antenna. Result was another good idea down the tubes.
I use it as sanity checks - it's quick, it's dirty, it works, and is accurate enough for a lot of real life problems. There's still a 4 inch diameter slide rule in the brief-case. Baffles the daylights out of the young'uns.
Orientation Angle Polarization Mismatch (dB) 0.0 (aligned) 0.0 15.0 0.3 30.0 1.3 45.0 3.0 60.0 6.0 75.0 12.0 90.0 (orthogonal) infinite (real world varies 20-35 dB) However, that's the polarization mismatch loss between different linear polarizations.
The loss between linear polarization and circular polarization is always -3dB whether it's vertical, horizontal, vertizontal, or horrible. However, that's only for a perfectly circular polarized antenna, which exists only in the manufacturers data sheets and college textbooks. In reality, the polarization is more eliptical than circular, which has considerably sifferent loss:
Axial Ratio or Min Polarization Max Polarization elipticity (dB) Loss (dB) Loss (dB) 0.00 3.01 3.01 0.25 2.89 3.14 0.50 2.77 3.27 0.75 2.65 3.40 1.00 2.54 3.54 1.50 2.33 3.83 2.00 2.12 4.12 3.00 1.77 4.77 4.00 1.46 5.46 5.00 1.19 6.19 10.00 0.41 10.41
Most TV and FM broadcast antennas are circular polarized. The home users have horizontally polarized yagis. The portable users have vertically polarized whips. Makes no sense, but what do I know. Examples of CP FM broadcast antennas.
I still use my 6" K&E. I never could get into the circular slide rules. However, this is easier.