Ask Slashdot: Why Does Wireless Gear Degrade Over Time?

acer123 writes "Lately I have replaced several home wireless routers because the signal strength has been found to be degraded. These devices, when new (2+ years ago) would cover an entire house. Over the years, the strength seems to decrease to a point where it might only cover one or two rooms. Of the three that I have replaced for friends, I have not found a common brand, age, etc. It just seems that after time, the signal strength decreases. I know that routers are cheap and easy to replace but I'm curious what actually causes this. I would have assumed that the components would either work or not work; we would either have a full signal or have no signal. I am not an electrical engineer and I can't find the answer online so I'm reaching out to you. Can someone explain how a transmitter can slowly go bad?"

Comments

Signal isn't chaning, the noise floor is

[Anonymous Coward]

As all of your neighbors add wireless routers, the noise floor goes up, and the usable signal goes down, even though the signal strength is the same.

Re:Signal isn't chaning, the noise floor is

[Migraineman]

More than likely, the older router was expecting a relatively clean RF environment, and was crippled when all the neighbors deployed APs nearby. The newer APs were designed to handle cluttered environments, and their more-advanced algorithms provide improved performance over the previous generations' products. As old equipment is replaced with new, you'll probably see the same degradation in performance until new countermeasures are developed (in the next gen equipment, of course.) Ref: arms race.

Re:Signal isn't chaning, the noise floor is

[Calos]

"Algorithms" aren't going to change because that requires a standard that must be followed by the transmitter and receiver. Unless s/he's upgrading from something like 802.11b to 802.11g, then there shouldn't be any such change. Possible exception would be a proprietary addition, but the problem remains.

It would be interesting to know if, when switching out the router, if s/he changed the frequency it's operating on. There are different bands that can be chosen even within the 802.11g spec, a newer router might have selected a less busy band automatically.

Then of course there's the fact that 802.11n completed changed frequency bands, from the 2.4 GHz region (which is extremely cluttered) to the 5 GHz region, which is relatively empty. That said, the higher frequency would be more impeded by solid barriers, e.g. walls. But it may compensate by higher transmit power, I don't know.

Hard to say if transmit power is really changing without being able to rule out other factors. But electronics do degrade. First suspect I'd think would be cheap capacitors. Poorly designed transistors could degrade, but this seems unlikely as RF band usually uses BJTs. Dust buildup could increase temperatures, which could hurt the efficiency and gain of these devices, but that's a rather long shot.

Re:Signal isn't chaning, the noise floor is

[currently_awake]

semiconductors are known to degrade if run hot. home routers don't have fans or heat sinks. budget devices often run hot to save on heat sinks and fans.

Re:Signal isn't chaning, the noise floor is

[sneakyimp]

I'd wager there are more algorithms involved than just the 802.11 protocol -- that protocol is the top layer in a stack of technology. Before you even get to the part where you are doing any kind of data handshaking, you might have a proprietary algorithm that filters your raw radio signal to weed out interference. There are also implementations of 802.11 on the market with non-standard features. Furthermore there is the inevitable forward march to ever-improving 802.11 standards. 802.11g is really old now. 802.11n is even old news. I've seen gobs of 802.11ac on sale at newegg.com and I don't even think the standard has been formalized. The "algorithms" are absolutely, definitely changing.

Re:Signal isn't chaning, the noise floor is

[blackicye]

That wouldn't explain why replacing the router fixes the problem, unless he just happens to be replacing the old router with one that just happens to have a stronger transmitter or better antenna. The pessimist in me says that the chances of that happening can't be 100% of the time.

From my experience with all the major manufacturers of consumer routers and switches, the problem is capacitor quality primarily.
Transmission range and stability will suffer over time because of unstable or insufficient voltage.

Also these devices get really hot these days. Most if not all are passively cooled, and don't even have much real ventilation in their casings.
They're designed to be cheap, and to last for at least 12 months generally.

I've replaced all the capacitors on several Linksys "Business class" Gigabit switch, they all started failing after about 14 months.
I did this to my own switch about 4 years ago, and it's still going strong today. I've also done this on an old Linksys WRT54G.

Re:Signal isn't chaning, the noise floor is

[bennomatic]

And now we all know exactly where you live ;)

Re:Signal isn't chaning, the noise floor is

[Anonymous Coward]

The FBI is not conducting surveillance today, which is why you don't see HotWorkoutPants. See you next week.

Re:Signal isn't chaning, the noise floor is

[zenith1111]

Then of course there's the fact that 802.11n completed changed frequency bands, from the 2.4 GHz region (which is extremely cluttered) to the 5 GHz region, which is relatively empty.

The better and more expensive Access Points can usually operate at both 2.4 and 5GHz, but most of the affordable 802.11n devices only operate at 2.4GHz (at least those I can find).

I was looking for a wireless card to enable an old laptop to connect to my 5GHz AP and I found that pretty much all of the cheap wireless cards also only operate at 2.4GHz.

Re:Signal isn't chaning, the noise floor is

[MSG]

PLEASE STOP OFFERING THIS ADVICE.

Increasing your WAP broadcast power does nothing to improve signal in the other direction, so while it will make your mobile devices show more bars, it won't actually improve network performance. TCP doesn't work unless a host can both send and receive (packets need to be ACKed), so even if the client receives further away from the WAP, it'll stop getting new packets if it can't notify the sender that those packets were received.

All that really happens when you increase broadcast power is an increase in interference with neighboring WAPs, which tends to lead other people to the conclusion that they also need to increase broadcast power in order to overcome the interference that you created.

Re:Signal isn't chaning, the noise floor is

[adri]

Hi. FreeBSD open source wireless developer here. I also work for a wireless company but this is all my own writing and is not endorsed or linked to my employer.

Don't do that. Let me repeat - don't increase TX power from what the card and regulatory limits say you can transmit.

Besides the regulatory limitations, the card may actually degrade if you increase the TX power. You may end up pulling more power than the card is designed or rated at. You may end up causing the output amplifiers to distort, which means you're not only breaking regulatory by spewing noise into adjacent channels, you're actually making your transmissions _worse_. It gets worse with higher transmission rates (especially 802.11n where the higher TX rates have much higher power density than the lower ones) - the Atheros driver implements per-rate TX power limits for this specific reason.

Chances are the manufacturer just has poor cooling, cheap part selection and all of that finely tuned RF front end is slowly degrading as a result. Buy an AP with better cooling or add better cooling yourself.

In fact, if you run the hardware at a _lower_ power output, you may find it lasts longer.

Re:Signal isn't chaning, the noise floor is

[Anonymous Coward]

Actually, radio standards (such as 802.11) never specify the receiver algorithm or architecture, as it is not necessary for interoperability. The receiver is only described in general terms.

Standards specify what is to be transmitted over-the-air, and might give a reference (typically in efficient) implementation of a transmitter for testing purposes. They might also give some examples of expected receiver output, also for testing.

Standards deliberately do not specify receivers, to give engineers an opportunity to innovate and provide differentiation in the market. Just one choice, for example, might be whether to do soft or hard decision decoding. (I happen to design receivers and transmitters for a living.)

As you've said, chances are any degradation are due to electrolytic capacitors drying out (and for that reason electrolytic are seeing less use), dirt and contamination, or maybe corrosion on connectors. Dropping output power might be one issue, but instability or inaccuracy developing in the timebase (crystal oscillator) is probably a bigger issue, and most actual failures end up being a power supply problem.

Re:Signal isn't chaning, the noise floor is

[needsomemoola]

There are 14 channels (frequencies) in 802.11b/g/n (2.4 GHz). There are only 3 that do not overlap though (1,6,11). The best thing to do if you plan to use the 2.4 GHz range is to run something like inSSIDer and see which of those 3 channels are least congested, then set your router to use that range. The problem with 2.4 is the lack of non-overlapping channels, and the fact that most routers have a default setting to pick the "least congested channel" but not conform to the 1,6,11 standard. So therefore you have all your neighbors congesting multiple channels by overlapping 1 and 6 or 6 and 11, using a channel in between. This is a nightmare for high-density areas (I do wireless for large conferences. It's a huge challenge).

In 802.11a/n (5 GHz), there are 23 channels you can use (depending on if you bond for N or not). This is like comparing a 3 lane highway to a 23 lane highway. Your density capacity is FAR higher than 2.4. The downside is that most mobile devices do not have 5 GHz radios, and 5 GHz, because of the nature of the higher frequency, does not penetrate (giggidy) as far as 2.4 (as you said). From a management point of view this is a benefit, but when you are trying to cover a large house it leads to weak signals at the edges of the house, if you center it. This is a good point to use multiple access points though (routers without the routing in lay terms).

So, the near future for wireless is 5 GHz (or, "Wireless A and N"). 5 GHz is catching on (iPhone 5 and a few Androids and Tablets have it now) and will start to get much busier, but the great thing about that is it's designed for it. The downside to the less-near future is that 802.11ac will add larger bonding in the 5 GHz range, which will lead to less available channels/capacity. We'll see how that goes...

One correction to your comment I have is that N is not 5 GHz specific, and it has not at all changed people from 2.4 to 5. It's only allowed more throughput than G or A did on those frequencies. It's caused more problems than not because of the fact it can use the 40 MHz window, taking up 1 of the 3 non-overlapping channels, pissing off your neighbors.

Re:Signal isn't chaning, the noise floor is

[pem]

No, silly, to provide stable power to the circuit trying to transmit or decode at 2.4GHz.

Noise (because of bad caps) from the power supply could easily cause jitter, which can reduce viable range considerably.

Re:Signal isn't chaning, the noise floor is

[gfxguy]

This is going to sound stupid, but I'm going to say it anyway - I'm not an electrical engineer, but my first wireless router to die died because the wall-wart power supply died. I don't imagine the electronics need all that much to work, but the transmitter might - is it possible to lose some power output over time? When I measured the output with my meter, it was like 3.2V, but it was supposed to put out 9V. I replaced it with a universal wall-wart and brought the router back to life, good as new.

Re:Signal isn't chaning, the noise floor is

[postbigbang]

It's not so simple. Let's review the history:

802.11a was first (there is another but obscure predecessor), and used 5Ghz. Advanced modulation techniques blew the doors off 802.11b @2.4ghz, which at 11mbit/s (on a good day), had a very low yield.

802.11g also uses 2.4ghz, but early products had trouble going back and forth between b and g, thus slowing throughput down, despite faster yields in g via advanced modulation techniques.

Enter N, was an advanced modulation scheme with higher throughput, first largely found on 2.4ghz. At first N was really fast, and had only a bit of trouble SLOWING DOWN for b and g. You have one collision domain unless you break out the b/g radio with the n radio. So, you have to dawdle while b goes through, then the channel is free again for something faster.

Then come the dual-band radios, and the dual band, dual radio routers which can walk and chew gum-- and handle paralellizing a 2.4ghz and a 5ghz conversation simultaneously-- major thruput.

The transceivers in the older routers appear to slow down, but in fact, they stay the same compared to newer ones for three reasons: 1) better firmware design that can switch back and forth quickly between protocols (where present) 2) have dual radios for bgn and (maybe a)Highband N and 3) the more recent the device, the more likely it has faster processing power inside the router. The final reason is that your backhaul might be getting faster without you knowing it; DSL gets faster but so also do cable broadband connections. And it's likely the driver in your machine is faster; they change all the time with small improvements, sometimes in real throughput.

Summary: the router didn't change, but newer stuff is faster given the same conditions for the reasons stated.

Re:Signal isn't chaning, the noise floor is

[RPI Geek]

No, we know.

Did the signal degrade, or the noise increase?

[SClitheroe]

Over 3 years I'd imagine a greater density of wifi devices all sharing the same spectrum to have appeared. Perhaps the signal level is the same, but the noise floor has increased substantially, degrading performance.

Re:Did the signal degrade, or the noise increase?

[pepsikid]

NO. If you use channel 2, then you're straddling channel 1 and 6, so you actually have to compete with more interference. Unless you live somewhere with no other wifi neighbors, like out in a desert or 3rd world country, never use anything but channels 1, 6, 11 or 14!

Re:Did the signal degrade, or the noise increase?

[girlintraining]

...but the noise floor has increased substantially, degrading performance.

Bingo! You hit the nail on the head. Wifi is now commonly found in most homes. And the overwhelming majority are b/g routers. That means that everyone's "last mile" internet is running on only three non-overlapping channels (in the United States), with a maximum capacity of only 54mbps for each of those channels. While your effective range decreases, your signal still continues to interfere with others out to its maximum range, which is typically around 300 feet. Beyond that, it's only a decibel or so above the noise floor (about -96dB) and is basically ambient. So consider urban density: In a 300 foot hemisphere, how many transmitters will be in that space?

Well, I live in a residential neighborhood that is mostly single-dwelling homes, which is about as ideal as you can get from a low-density city environment. Using a pringles can, I took a neighborhood survey and found about 26 access points within 300 feet of my home. Now, this is a survey that took several days to complete because of the marginal signal integrity, after which I drove my car in circles matching associated clients to those APs. Each access point had approximately 2.25 clients associated with it. So that's about 60 transmitting devices, in an ideal urban environment. And that's just those using wifi.

2.4GHz is also used by: Wireless phones, microwaves, wireless "hifi" stereo systems, etc. It's also used by wireless mice/trackballs and keyboards. So, realistically, I've got at least 100 devices that are transmitting with a signal high enough to interfere with the front-end RF of my wifi.

Shannon's Law stands tall in all of this: As you increase the noise floor, the amount of data you can transmit regardless of encoding scheme or receiver selectivity falls proportionally. Every device added decreases your own devices' performance.

Solutions
I found that by setting my router to 'g' only and then forcing the bitrate down to 24mbps, I was able to get a much more reliable and speedy signal. Every WiFi standard is designed to cope with interference by renegotiating to a higher or lower bitrate dynamically. Which would be fine if they were isolated, but in an environment where they're in close proximity to each other, what happens is as each device broadcasts and interferes with the other, they detect this and then renegotiate, generating more interference; And pretty soon you've got routers constantly in a state of renegotiation, with fluxuating bitrates. Manually force your router to a specific bitrate and don't allow re-negotiation, and you'll find that those momentary spikes in the noise level won't wash out your signal -- renegotiation takes 10--30ms, and during that time, you can't send/receive any data. The data burst that caused it is over long before the renegotiation completes.

So in short, it's not your transmitter, it's the environment. Take your transmitter out of its default settings and enable RTS/CTS (if available) and you'll be fine. Another, more sociopathic answer, is to get a 100W 2.4ghz booster (you'll have to build it), mount it on your roof, tune it to one of the 3 non-overlapping channels (I suggest 1 or 11, since most microwave ovens tend to tune at the middle of the band -- channel 6), and then let it run for about 3--5 days. Everyone will bail off that channel because nothing tuned to it will operate over a distance of even a few feet. Again, very illegal, very sociopathic... but very effective. You'll have to do "plow the spectrum" about once every month or two, so count on downtime.

Re:Did the signal degrade, or the noise increase?

[TheGratefulNet]

chinese capacitors.

not kidding. newer gear uses junk parts that the vendor or builder decided to use instead of brand name trustable parts. or, they sought out real ones but got fakes. in electrolytics, its a mostly fakes world ;(

they last a year or a few years, tops. you'll see the cans bulge and burst at the expansion caps, at the top (alum creased areas).

if its on the digital side, you lose all funct.

if its on the analog side (rf, etc) then things can degrade before fully failing. I've seen this in audio gear, too, btw.

cure is to buy known good caps from known vendors (digikey, mouser, newark, etc) and install them yourself. get a hakko desoldering tool, pull out ALL electrolytics and get same LS (lead spacing size) and value caps. try to increase the voltage on them (their rated voltage) as the vendor often gets that part, wrong, in safety margins.

usually, its the power supply that goes bad. and usually its the caps. if you replace your caps, you can convert a $50 consumer throw-away into a $5000 enterprise class gear that will actually out-run most commercial gear simply by using GOOD low ESR caps instead of fakes that almost every one ships with.

panasonic, nichicon, others make good low-esr filter caps. they are a dollar or so, in price. not expensive. not hard to replace.

swap them now or wait for a failure. either way, this is almost always the cause of networking and computer gear these days.

Re:Did the signal degrade, or the noise increase?

[girlintraining]

I'm actually in a suburban city in the midwestern United States, and the neighborhood is very middle class. If I go downtown, there are over 250 access points and well over 1,500 clients visible using the same method of discovery. If I head to the 'rich' part of town, there are fewer access points, but a lot more clients connected to them on average. And the poor parts of town are usually apartment housing, which is much denser and so has a higher number of access points... but a lower number of associated clients per, on average. So overall, except for high density commercial environments, the number of devices per mile doesn't vary much based on income -- it's pretty locked in on the number of people per mile. But again, these are averages from having sampled dozens of neighborhoods. In any given neighborhood, you're going to find locations that are very dense and others almost barren, sometimes separated by only a few hundred feet. Statistics, you know...

I would like to know more about how you measured the number of associated clients though.

I cheated, and it was probably illegal, though petty. I salvaged a grey fiberglass enclosure pod previously used for housing cable TV equipment, and then modified it so it could be quickly and discreetly attached to a telephone pole. Inside was a 'microATX' computer with a pair of SD cards to boot off of, and was underclocked and undervolted, the fans stripped off and replaced with large passive fins. water-resistant wire netting was added along the inside (the part facing the pole) and thermally bonded to oversized heatsinks. Two wifi antennas were mounted along the front at 45 degree angles (to catch both vertically and horizontally-aligned signals) and attached via SMC connectors to adapters. One was 'a', the other was 'a/b/g/n'. Last was a special diagnostic adapter with three antennas which cost an arm and a leg. It was a spectrum analyzer designed to identify sources of high EMR and localize the source. The remaining space was packed tight with high capacity deep cycle batteries. The unit could run for about 86 hours before it ran out of juice.

It recorded every MAC address and associated BSSID, etc., as well as encoding and some other properties. You can catch new clients because before they associate, they broadcast the SSID they want to connect to in the clear. Over a period of three days it collected all that information and then wrote it out to the SDcards. Drag it home, hook it up to the car charger and pull the cards. Rinse, wash, repeat.

You're the first post I've come to on this thread that has mentioned the sad relationship between microwave ovens and channel 6. Other people mentioned the noise floor early in the discussion, but I didn't notice any of them actually referencing that to Shannon's Law, either. You should be modded informative a couple of times, but I won't be surprised if the 'sociopathic answer' part gets you downmodded instead.

I wouldn't be surprised either. Slashdot ought to just replace their mod system with "like" and "dislike", since that's really what it boils down to. I frequently get downmodded for saying something that is technically correct or possible, without addressing the ethical considerations. I prefer to tell people the whole truth and let them make their own choices, rather than leaving out critical information because I don't trust them enough not to do something stupid. People need to know what's possible, not just what's legal or ethical. This idea was summed up beautifully for me years ago on a forum far, far, away, when someone wrote "You can't expect a terrorist to care that his car bomb is taking up two parking spaces." Criminals don't give a damn about you or the law; They are there to grab the low-hanging fruit, and unless you know what's possible, how they operate, you can't defend against attacks.

You can't be a good white hat without having worn the black hat.

Re:Did the signal degrade, or the noise increase?

[TheGratefulNet]

Everything electronic degrades over time. The only electronics we have that are mostly proof against it are in orbit and cost exorbitant amounts of money.

it didn't used to be that way.

a time-travel into the past, via ebay, can be insightful. over the last year or so, I've been collecting test gear from the 50's 60' and 70's (from good brands, fluke, tek, HP back before agilent, etc). one brand that really impressed me was 'power designs'. search for it with the word 'precision' and you'll see samples of what was built tens of years ago (some half a century!) and yet they still hold their values to high tolerances, more so than quite a lot of 'high end' supplies made by high end brands today! I bought quite a few, tested them (after cleaning them up) and they showed amazing stability, down to microvolts. the switches were good, the meter was good, paint and metal was fine, wiring and circuit boards were fine. even 40 yr old caps were fine, to be honest. I replaced the caps as a matter of course, but I could probably get more time from the gear if I wanted to.

these costed a few hundred dollars back then (30-40 yrs ago). now, you can't even buy that quality (and the ebay sourced PDI precision gear tends to go for $100 level prices, today). but back then, it was expensive but not outrageous. and it lasted more than many peoples' careers! the stuff STILL runs! and stays in spec.

most people have no exposure to this. all they know is what they buy at frys or best buy or worse, ebay china stuff. they *assume* that things won't last more than a few years, usually less than 5. they're OK with junking it and starting with new gear! ;(

older gear was meant to be repairable and last for decades. my test lab is mostly filled with vintage gear that was trivial to restore and would cost me 10x or more to get new equiv stuff, if even possible.

look at 'general radio' and search for pics of what their manual switched voltage dividers look like. I have one that is 50 yrs old and can't be matched with even today's gear.

things USED to be built to last. they really did. and people who are in the 20's and 30's have no idea about this strange old idea, either. that's the pity of it all.

cheap electroytic capacitors

[pentabular]

..have a tendency to degrade and fail over time.

Re:cheap electroytic capacitors

[Theaetetus]

>cheap electroytic capacitors have a tendency to degrade and fail over time.

Not significantly over 2 years and you don't use electrolytics in the in IF/RF signal path in a 2.4 & 5.8GHz radios.

True, but you do use them in your cheap switch-mode power supply, and as they degrade, you get additional AC noise on the rails of your amplifiers that are in the IF/RF signal path. Particularly in cheap routers that are operating near the limits of their amplifiers, voltage drops on the rails could cause clipping of the high frequency signal, which will result in dropped packets, required rebroadcasting, etc.

Re:cheap electroytic capacitors

[artor3]

I don't think so. I'm an RF electronics engineer, and we do all sorts of accelerated stress testing to check out the second half of the bathtub curve, and I've never seen degradation on anything in the RF/IF path inside the chip. The typical "wear out" failure mode is cracks forming in the protective seals around the chip and letting in moisture (from the air), which causes leakage (high current draw) and eventually just shorts out something important and the chip dies. At any rate, that shouldn't be happening in just a couple years, unless the submitter moves his PC back and forth between the freezer and the sauna every week.

One possible explanation would be crystal aging. RF equipment tends to rely on extremely accurate quartz crystals to provide reference frequencies. Those crystals tend to drift as they age. If the design was already near the edge of acceptable frequencies, an extra 10 or 20 ppm from aging could easily result in several dB of degradation.

Another poster pointed out the possibility of the router using a crappy switched-mode power supply, which is also a good explanation. I would hope that they would power the RF chip through a linear regulator with good noise rejection, but who knows? That sort of switching noise can absolutely interfere with the radio's performance.

analog transistors age

[jfb2252]

This is a hypothesis based on peripheral involvement with analog and digital RF at 0.5 and 1.5 GHz for twenty years.

AFAIK, the output stage of anything broadcasting above about 2 GHz has to be analog, with the lower frequency signal mixed into a carrier at the higher frequency. Digital synthesizers and chips which can deal with 1.5 GHz directly are still very expensive and are unlikely to be used in the consumer routers. So the final output stage is likely an analog RF transistor.

Analog transistors change characteristics with age at elevated temperature, where elevated is anything over 20C. Implanted ions diffuse with time and temperature, changing junction characteristics. The small structures required by high frequencies are more sensitive to such things.

Power adaptors to blame.

[Anonymous Coward]

In my experience power adaptor degradation is the main culprit. Over time the adaptor will provide lower voltages and a less stable current. This translates into a lower signal output and higher noise respectably. I've seen bad adaptor turn repeaters into signal jammers - trust me, that was not an easy issue to troubleshoot...

Several causes, but a few that spring to mind...

[Tastecicles]

1. slow burnout of emitter gear due to thermal degradation (yes, clock chips and transistors get hot, as do solder tracks and joints). Thermal runaway can occur if a solder joint fails and arcs, or overvoltage causes signal tracks to vapourise.
2. ionising radiation, particularly on unshielded components such as antenna conductors (I've seen something like this occur on an externally mounted amateur radio antenna: the sunward side of the antenna completely degraded, the result being that the only signals received (or sent) were on the shadow side).
3. component quality on consumer gear is not as stringent as it could be. Components can and do fail, and considering the number of components in a lot of consumer gear, it's a wonder any of it actually leaves the factory.
4. the noise floor of several years ago was far, far lower than it is now. The ERP of newer gear is (by design or by necessity) higher than older gear as more and more transmitters have to share the band. As a result, the signal quality taking a dive may be at least partly illusory. The equipment may actually be perfectly fine.
5. parasitic structures in semiconductor packages may be the catalyst for failure, either immediate or delayed. Such structures may be as small as a single atom of chlorine embedded in a crystal of germanium - innocuous at first (undetectable, even), but over time and use, that contamination will alter the chemistry of the semiconductor, possibly causing it to bond with the package material and rendering it useless. This might not even be an issue in high powered gear like regulators but in something like a microprocessor, it's a showstopper.

Re:The Hamsters get tired

[Alsee]

The transmitter range isn't decreasing.

It's actually due to the expansion of the universe. It's because your house is getting bigger. You just don't notice it because you are expanding at the same rate. Try going on a diet.

-