>Steve Baker prefers a point to multi-point microwave approach,
I'm no expert, but I would have thought a Line Of Sight approach naturally fits lampposts, since from one lamppost you can usually see others.
A new design of mobile cell site that sits on top of an existing street light has been developed British design company The Technology Partnership (TTP). Street lights are an obvious place to put mobile network infrastructure. They're erected in the places where there are lots of people and they have a power supply. While …
Around here there's an ongoing drive to turn off street lights between midnight and 5 am as the council's skint.
Hopefully that means each individual lamp is controlled so it may be switched on/off remotely without killing the power supply. Knowing how our council likes to "plan" things they probably just pull the plug at the control centre. That's the kind of thing that could kill this idea dead.
Can't imagine residents being happy about loss of mobile signal overnight. Sure, I'd hope there'd be fallback to a larger cell, but I can easily see the cellco's removing the larger cells if they have street-light coverage. Cost savings and all that.
We had the same. A Yorkshire company made replacement photocells that worked out midnight from dawn/dusk observations, calculated the date so it could adjust to daylight saving times, then turned off from midnight to 6am, all independently of any external input in a package that could replace a standard dawn/dusk module. And a mere 6 times the price of the standard photocell.
But they used a PIC microcontroller running some sequence they'd written which has the following "features". At the equinoxes, the controller had a divide by zero error which crashed the controller and turned the lights on during the day and off all night until the power supply was flipped at the roadside pillar. A really, really dark cloudy day, around 4pm in March or October, which shifted the dusk detection time by more than half an hour, caused the system to recalculate the date and recalibrate, meaning it failed-safe to lamp on all the time. The system has an internal battery maintained clock which never runs at the right rate.
To correct these software bugs, they had to replace every single one of the 64,000 they'd installed throughout the county. And the new ones STILL go wrong at Equinox.
My solution? A free-running counter, any reasonable speed would do, a photocell, a divide by two stage, a second counter connected to the divide by two stage, a latch, a comparator and a 555 delay circuit. At dusk, turn the lamp on and start the counters running at full speed and half speed. If the running counter goes above the value held in the latch that being the value of the half speed counter from the previous day at dawn, the light goes off, triggering a time delay of around 6 hours to hold the lamp off unless dawn occurs before then. The counters run on until dawn, ready for the next day.
No microprocessors. Nothing expensive. No software to go wrong, just a true midnight, self-calibrating system. Who needs 0.1% accuracy of timing on a street light, FFS? Most appropriate solution at the best price.
"My solution? A free-running counter, any reasonable speed would do, a photocell, a divide by two stage, a second counter connected to the divide by two stage, a latch, a comparator and a 555 delay circuit"
Still overly complex and people HATE dark streets.
Dimming 90% and brightening up when movement is detected seems to be the way forward. Of course the "quality british designs" (meaning the idea is ok but the execution is crap) will win out thanks to "buy UK" campaigns, instead of picking what works and actually saves money.
BTW, the electrical cost savings in turning off non-led lamps for part of the night are more than made up for by thr added labour and hardware costs caused by premature failure thanks to the things being cycled twice as often as they were originally intended to be (discharge lamps have lifecycles measured in hours OR cycles, in a similar manner to pressurised aircraft - thankfully an over-cycled lamp doesn't blow its roof off midway through the night)
"Powerline ethernet? "
Discharge lamps are wideband RF emitters (just like powerline ethernet).
The discharge lamp's emissions will render the powerline Ethernet useless.
LED streetlamps? Don't know for sure, but I have heard lots of stories that some switched mode power supplies used with domestic LEDs emit lots of wideband RF. So same problem as discharge lamps.
"It’s all very well being able to get the mobile signal to the eNodeB on the street light but getting the voice or data from the lights to the phone network is a tougher problem.
TTP kind of employs a SEP (somebody else’s problem) field solution to this, but offers a partial answer in the form of an a Gigabit interface for S1/X2 and provides Power Over Ethernet..."
And there's the elephant in the room. It still requires digging up the roads/pavements, or persuading existing infrastructure owners to permit access to their ducting. So massive, widespread, long-term disruption with lamp post cells, instead of localised short-term disruption from building a bigger cell tower with more coverage.
This misses the fundamental point! No need to dig up pavements. Gigabit Ethernet isn't cabled all the way back to the network core :-) There's just a short cable run down to the wireless backhaul node which is clamped to the post. And power is supplied from the cell to that backhaul node using POE.
This post has been deleted by its author
Small cells are (largely) used fpr capacity not coverage. You get the ectra capacity by re-using frequencies/codes much mpre pften because small cells cover...a smaller area which mavro cells can't.
Cost is a big issue too,small cells are an order of magnitude smaller than macros.
" localised short-term disruption from building a bigger cell tower with more coverage."
The point about the bigger tower with more coverage is that it covers TOO MANY phones. Telcos only shrink cells when they have to. A smaller one costs just as much at the backend as a larger one.
Question - if you're extending your 4G coverage, say, by using these devices, and they themselves connect back over 4G... isn't there a problem here?
One device fails near the main 4G tower, and the entire street/town all fall over?
And if you already have 4G reception to the streetlights.... why are you adding more 4G?
And if you have to do some other backhaul to get to these 4G cells, couldn't you have just done that anyway - because presumably that backhaul ends up somewhere nearby (a BT cabinet or similar) and you could have just mounted a pole or cell on the back of that, no?
It all seems a bit pointless compared to just slapping into another mobile tower. Maybe relevant in extreme rural areas where you're not allowed to even erect a pole but, then, you were allowed to erect streetlights so presumably a pole of the same height would have the same permissions, no?
It very much seems a solution in need of multiple problems.
I am no expert, but as I understand it the advantage of using numerous smaller cells allows the same frequencies to be reused by different cells at the same time, making more bandwidth available than would be possible with a single large cell covering the same area. The spanner in the works is the backhaul, as linking to a larger cell or daisy chaining through the small cells negates the bandwidth advantage that was created.
Not necessarily. The point is to reduce the collision domain. It's a little like the queue buffers in a switch/router, if data-in temporarily exceeds output capacity, you can store the traffic and send it in an efficient stream, taking advantage of its (hopefully) bursty nature.
Of course, wires are best, so only use this stuff for devices that need it. That means, stop pushing "wireless internet" for home use. Dig up the roads and lay the fibre! There are good reasons why we don't use wireless networking in data-centres.
STREET LIGHTS SET TO BECOME MOBILE MASTS
12:07Tuesday 23 December 2003
Tuesday, December 23: Lampposts all over Hemel Hempstead may soon not be quite what they seem.
T-Mobile put in a clutch of planning applications this week for the removal of several lamp standards, to replace them with look-alike telephone masts for the new 3G system for mobile phones.
The masqueraders are planned to be placed on footpaths outside Greenacres Tavern in Bennetts End, at the junction of Eastbrook Way and Adeyfield Road and on land adjacent to Long Chaulden at the junction with Pulleys Lane.
Same approach was used in Metricom's Ricochet wireless system.. 20 years ago! I wonder who's patents expired..). The smaller cells and reduced transmitter receiver distance pay huge dividends in density and bandwidth, and can allow the clients to lower transmit power, lowering the noise floor for other users.
The decision to keep the back-haul mix and match is GOOD thing, not a tap out. It means the same box can be adapted in a flexible build that uses different back-haul strategies where appropriate, and allows the in place upgrade of components.
Ricochet used to bounce the signal through a couple of peer nodes till it got back to on with a Hard link. In this case, a directional microwave antenna may serve to get your signal down the block to the nearest street level junction box, then a short hop over to the fiber network.
Of course in fantasy land, we should put an MIMO AC access point and Micro Cell on each pole, with dedicated fiber links for each node running to a Municipal fiber build.
I just hope they have good security on those POE ports, otherwise you will have to look for suspicious pass through modules every time you make a call.
I was a Ricochet customer - still have the modem around somewhere. Pretty cool at the time, and the lightposts where I live still have the Ricochet hardware hanging - mostly unplugged from the light sensor. Guess the municipalities never thought about what would happen if they went bust.
There are still a few Ricochet repeaters hanging from streetlights around my neck of the woods. Thinking about it, Ricochet was a forerunner to the Light-Squared debacle, as Ricochet's business plan relied on using unlicensed spectrum - albeit LightSquared's spectrum was licensed, it wasn't licensed for what they wanted to use it for.
and the tome's that just keep giving..
" SEP (somebody else’s problem)"
aka how to make mountains invisible :)
For anyone who hasn't yet read the 6th volume of the trilogy (not by DA obviously) should give it a go. It tries a little hard in the first few pages and overdoes the swearing a bit, but once it settles down is practically indistinguishable (sp?) from DA's prose. So much so I did actually laugh out loud several times whilst reading it - highly recommended
I would have thought a hybrid solution, using fibre to connect main nodes, as fibre can be run parallel to the mains cabling without issue, they could maybe share ducts.
Then power-line networking over the existing power lines from the main nodes to adjacent lamp post tributary nodes. May need isolation transformers (not sure what the efficiency loss would be) to isolate mains segments to prevent CDMA collisions killing throughput, there also be an issue with the lamp posts acting as RF antenna and leaking interference all over the place.
Noisy ballasts on the lamps may cause interference, but they could upgrade the lamp to a nice efficient low-noise LED unit while they are installing the cell.
I am sure there are things I have not thought of, I am not an expert in street lighting.
Biting the hand that feeds IT © 1998–2022