Looking for equipment to improve rural 4G signal reception and speed

RouterOS Wireless Networking
1

Hello,

I am looking for a way to improve my 4G speed in a remote rural area.
I am not an expert on wireless/mobile networks.

Currently I am using Huawei B311-221 and the LAN clients are a few WiFi and Ethernet connected devices.

Normally, when connected to the 4G network, that router connects to one of 3 cell towers. For each cell I can read in its web interface:

Cell 1 (gives me ~30 Mbps download and 1 Mbps upload speed):

CQI: CQI0:5 CQI1:8
RSRQ: -7.0dB
RSRP: -117dBm
RSSI: -89dBm
SINR: 4dB
4G uplink frequency: 17700
4G downlink frequency: 18650
Band: 3

Cell 2 (gives me 6 Mbps download and 1 Mbps upload speed):

CQI: CQI0:8 CQI1:127
RSRQ: -10.0dB
RSRP: -109dBm
RSSI: -87dBm
SINR: 0dB
4G uplink frequency: 8866
4G downlink frequency: 9316
Band: 8

Cell 3 (gives me < 100 Kbps speeds):

CQI: CQI0:5 CQI1:127
RSRQ: -16.0dB
RSRP: -111dBm
RSSI: -87dBm
SINR: -8dB

The ISP claims to provide about 110 Mbps download and 35 Mbps upload speeds - obviously I am quite far from that and I am not happy with it.

I talked to my ISP and they explained that the speeds I get in this rural area (30 Mbps download and 1 Mbps upload when it is “good” and 7 Mbps DL and below 1 Mbps UL when “bad”) are “fairly normal for such region” but that is not what I pay for.

Here is a short reconstruction of the discussion:

ISP: The cell tower is configured so that it prioritizes the voice calls over Internet connections. So, if the cell is loaded with voice calls, your speed may drop and you may also be switched to another cell tower.

Me: Can I use better equipment with directional antenna?

ISP: There is no specific direction, the signal is the same in all directions but you could give it a try. It is a good idea but an investment too.

I am OK to invest in equipment but I want the most suitable one which will give best result, so I digged deeper.

Mikrotik’s support suggested that I look at the LHG product line and use a directional antenna (as I thought initially). Obviously, the best product is ATL LTE18 kit.

Then I had another lengthy discussion with my ISP. As mentioned, their explanation about prioritization remains: voice calls are #1 priority, Internet access comes next. Some details from that discussion:

Me: Suppose I use a directional antenna as on. What should I look for primarily:

  • 2x2 or 4x4 MIMO
  • high dBi
  • higher LTE category

Which is most important?

ISP: 4x4 MIMO and high dBi directional antenna. LTE category is less important.

Me: AFAIK, cell communication works this way: the mobile device connects to the cell with best signal. Is that right?

ISP: Yes.

Me: Are you saying that in the case of 4G Internet I pay for, it is not exactly like that but I may be switched to a cell with worse signal because your prioritization system may need the better cell for voice calls?

ISP: Correct.

Me: So, suppose I buy that good antenna and router, then I point it to the cell tower which gives me best speed. Then, your prioritization system switches me to another cell. What will happen then?

ISP: You will be experiencing a very bad signal reception and very low speed.

Me: Then what’s the point of using a directional antenna as you recommended? Doesn’t that refute the whole idea?

ISP: Indeed, it is not a perfect solution as the antenna will not auto rotate to the other cell.

Me: What if I take my current router and my laptop and go right in front of the cell for the sake of testing?

ISP: The result would depend on the cell load but generally you would experience the full speed as per contract.

Me: Why do I experience constantly bad upload speed on all cells?

ISP: The information we see here during the tests we made is that the reason is the bad signal reception on your side.

Mikrotik’s support explained that Router OS supports cell locking but the info says:

It is possible to lock R11e-LTE, R11e-LTE6 and R11e-4G modems and equipped devices to exact LTE tower.

i.e. nothing about R11mL-EG18-EA - the modem of LHGG and ATL LTE18 kits, only about other modems. Support also mentioned that it is possible to use two directional antennas at the same time with two uplinks but refused to elaborate further, directing me to the forum.

My concern with that is: if cell locking is possible with the equipment I buy and I lock to the best cell, then my ISP’s prioritization switches me to another cell, I will be disconnected with an antenna pointed towards a cell which refuses connection, OR if I use automatic switch-back to the good cell there will be constant switching back and forth between the preferred cell and another one. OR… I don’t understand a thing about how this works.

Based on all this, I am quite confused and I would like to ask:

1. Is my ISP lying to me?

I am asking this because although I understand the technical logic of the load balancing they explain, it makes no sense. As a mental exercise taking things to extremity: Suppose I am right in front of the cell tower. Would it switch me to a cell in another city (which I can’t even reach), just because of “prioritization”? To my (non-expert) mind that contradicts the whole idea of cell communication. Additionally, it is quite unfair because I pay 10 times more for that Internet plan compared to someone who pays “just for voice calls” but he has higher priority (but I am OK to ignore even that if there is a technical solution).

2. What is the proper technical solution and equipment for the described situation?

I would like to have the best possible speed.

P.S. Sorry for the lengthy post. I wanted to give full details, so that my questions make sense. I hope that is OK.

2

The biggest problem in your discussion with your MNO/ISP is that most likely the support guy doesn’t know much about technical details of how LTE network works (in general and their network in particular). I know a bit, I worked for 17 years for MNO as (senior) radio engineer dealing with GSM, UMTS and LTE.

An example: network will never just “switch modem to another cell”, modem first has to measure the “another cell” signal strength and report it to base station. All of cells have really low signal strengths (RSRP lower than -100 is bad and lower than -110 is shitty), but if they come from different directions, then directional antenna will improve signal strength from one direction and degrade others. Then baee station may decide to “switch modem”, depending on MNO’s strategy. Setting “cell lock” on modem actually prevents modem from reporting measurements for other cells so network won’t switch modem to another cell. But this also impacts CA which is most often not desired.

The problem with uplink speed is very likely due to combination of low signal strength (which indicates big path loss) and low Tx power of your modem (base stations transmit with powers between 20W and 100W while modems and phones are limited to 0.2W), using directikbal antenna will improve that as well.

In such bad conditions, MIMO order higher than 2x2 (which is minimum required) won’t help at all.

Regarding gain: one of cells mentioned operates in 900MHz band (band 8 was traditionally used for GSM in Europe, in LTE usually provides low capacity) and the other one in 1800MHz band (used as capacity extension for GSM, in LTE often provides decent capacity). So it’s important to find solution which has good (or at least decent) performance in both low and mid bands. When antenna gain is characterized only by a single number, it’s typically true for higest supported frequencies. The figure for low band is almost always lower and for certain antennae it’s really bad (some MT products included).

3

Thank you for this information. It is very welcome for a non-expert like me.

My MNO/ISP’s support is (softly speaking) very bad. The discussion(s) I shared are “the best” info I was able to get. Unfortunately, there isn’t much of a choice of ISPs in this region.

The basis of what they said was “tests” in which they required me to generate download and upload traffic, so they can “measure the connection quality”, “trace” and I don’t know what else. I don’t know what they actually measured, because they didn’t tell me anything more than what I already know - download/upload speed and signal strength. I think they have some way to access the router without me authorizing that. I suppose it is a deliberate backdoor because they supply the routers as part of the service. Fortunately, it is possible to use one’s own device, which is why I am researching.

Could you please explain why they asked me to generate traffic?
Also what is “trace”? (I know what traceroute is)

Regarding signal strengths:

Using an Android phone I walked around checking the signal menu (which is obviously very limited but still something). The best parameters I could get were on the very roof of the building: -83 dBm and 10 asu (compare this to the -110 dBm and 3 asu right next to the router inside the room it currently is). Unfortunately, I can’t put the router on the roof because it is not weather proof but I wanted to share this additional info, so we can talk about possibilities. I could put the router temporarily on the upper floor (-97 dBm) which gave me stable connection to “Cell 1” with about 30-35 Mbps download and (still) 1 Mbps upload speed.

I wonder if using some more powerful antenna right on the roof (where I can mount it on a pole on the chimney) might improve things and how much.

What do you say?

Any thoughts, clarifications and specific recommendations from an expert like you are very welcome.

4

Using a mast on the roof might provide you with a decent line-of-sight to the nearest tower that will possible allow you to utilize Carrier Aggregation (CA). In such scenario, the LHG LTE6 kit might be a suitable option.

We’ve got some customers in the archipelago who have a line-of-sight of approx 5-6 km to the nearest tower. They achieve download speeds ranging from 40-60 Mbps using LTE and around 80-100 Mbps with NR (5G).

5

I can’t say for sure what would they do while you would do the “tests”. However, most mobile networks elements (switching servers, routers, base stations, etc.) do come with debuging facilities and it is possible to “trace” communication of a subscriber (based on IMSI number, i.e. SIM card) through those elements. And to see the network’s view on communication between modem and network. Most of available information is identical to what would advanced modem management show (but it’s not available in most mass-produced modems) as it’s signalling communication between network and modem (and if it does work two-way, it’s seen the same on both ends). And then there’s some information only available on one end because one end collects information and uses it to make decissions but doesn’t send it to the other end. Because most control of modem-network connection is done by network, modem sends more information to network than the other way around (e.g. it sends “measurement reports”, which include all cells measured whose signal strength and quality exceed set thresholds … if serving cell strength and/or quality drop bellow set thresholds. Based on measurement reports betwork “switches modem” to another cell while decission algorithm and parameters is not known to modem, MNO OTOH knows these as they are operator setable and knowing them does help zo understand why modem behaves in certain way in certain conditions). So proper “trace” from network side can reveal a lot of information. There are literally thousands of parameters which can be set by MNO, so it’s impossible to fully understand behaviour simply by observing behaviour in a few use cases. Networks (or even parts of same network) are not comparable between each other and recipe which works in one particular case may not work in another (seemingly almost identical) case.

Regarding signal strengths: it’s true that different devices will show different signal strengths in the same spot. Mostly this depends on design of internal antennae, some are real gems and some are mediocre. When it comes to reported signal strength they should rrport similar values but often they don’t because e.g. antenna gain is not constant over whole fequency range but software uses single number to do tge naths. Also sensitivity of receiver is important, linearity of input power amplifier as well. The last two affect also actual modem performance (in receiving part) while imprecise antenna gain value is more or less cosmetics.
But whatever the imperfections, higger signal strength is always better (as long as it’s not too high which can cause receiver overload and decrease in performance, but that realistically only happens right in fron of base station antenna). So whatever antenna is attached to whatever device, try to position and orient antenna (or device if internal antenna is used) so that signal strength is the best.
This also helps with Tx performance. As I already mentioned, Tx power of mobile devices is around 2 order of magnitudes lower than of base station. Which would mean correspondingly lower throughput if it wasn’t for better Rx sensitivity of base stations, which mostly compenstates lack of Tx power of mobile devices, but not entirely. There are MNOs who, trying to increase signal coverage (and to show more “signal bars” on mobiles), bump up Tx power of base stations … which does improve downlink performance, but makes uplink communication almkst impissibke in regions with low signal strengths (which is where you are). Using high-gain antenba on modem side also increases uplink signal and that almost always improves uplink (and improvement is most prominent in case of poor to mediocre signal strength, I’m affraid you’re at lower edge of usable signal strength).

I’d say you should try to improve your LTE experience by using decent directional antenna set (it should include two antennae for 2x2 MIMO), but don’t expect miracles. There’s also (small, but real) possibility that even if using good antennae your service won’t improve.

6

Mkx, great sum-up. You should definitely try to write a LTE HOW-TO guide considering your pedagogical skills!

7

You can also check https://www.cellmapper.net/map to see where the nearby towers are located (not exact/complete but worth checking). One thing a directional antenna (pointed at a tower) often helps with is the upload, which seems to be the bigger problem here.

Since 1800Mhz looks best from your data, even a SXT-LTE might work and that might be easier to tweak/test than a LHG (e.g. you can point it out a window more easily to do some tests).

8

Thank you everyone!

Please let me check if I understand correctly:

I am looking at:

LHGG LTE6 kit (LTE Category 6, no MIMO info, 17 dBi antenna)
LHG LTE18 kit (LTE Category 18, 2x2 MIMO, 17 dBi antenna)
ATL LTE18 kit (LTE Category 18, 4x4 MIMO, no dBi info about antenna)

FWIW, when I asked support about the missing details and the info I received was:

LHGG LTE6 kit - mimo 2x2 is supported + ATL LTE18 same antenna gain as LHG-18 series. Ultimately, ATL series is our new approach, hardwarevice devices are simmilar, but ATL series, tends to focus more on the speed, LHG with its honeycomb design antenna, focused more on extending signal range.

If I may accent again, I don’t see any info about cell locking capabilities on ATL LTE18 kits, as mentioned in the initial post. So, any info about this would be very welcome as Mikrotik support refuses to share further details.

Based on everything so far my questions are:

  1. Is there a point to invest in a 4x4 MIMO antenna at all? Or it would give the same result as 2x2 MIMO in my particular case? (+ How does one know all that?)
  2. I understand this may be a blasphemy to ask in this forum but still (purely technically, zero marketing): Is Mikrotik’s equipment the best for the particular case or is there something even better? (and which is it?)
  3. Or are my efforts futile?
  4. @mkx can you recommend any good educational materials on the subject? (books, tutorials, etc)

Any other info is very welcome! I hope my efforts are not futile and that there is a possible solution.

9

My 5 cents:

  1. Not at all. As already mentioned, higher MIMO orders are helpful when signal levels are at least decent. Also base stations have to be 4x4 MIMO and I’ve so far only seen 4x4 implemented on higher frequencies (e.g. 2.6GHz and higher). In your case I’m 99.9% sure it would be just waste of money
  2. my personal take is that MT equipment is more or less medium. Modems seems to be slightly outdated and their antennae seem to be modest at best in low frequency range … this is hard to verify as MT doesn’t publish good information about characteristics of their antennae. Some other (antenna) vendors are much more open than this (one can always have their doubts about how realistic is that data though), example is Iskra Antene catalogue
  3. you’ll know after you try your best. You just have to decide whether investing a few hundred euros worth into gear which may prove wasted money is something you can afford. My far-fetched feeling is that you should be able to improve things to certain extent, but as I wrote, don’t expect miracles, with that signal strengths you’ll never be able to get anywhere near maximum throughput of cells covering your location (which might be somewhere near 250Mbps/100Mbps with fully working CA and high-category modem).
  4. Actually I can’t … my (incomplete) knowledge comes from numerous books, numerous courses (most taken, some given with useful feedback from professionals attending), some confidential documentation from different network and phone manufacturers (NDAs signed) and many years of hands-on experience.
10

Thank you. Very valuable info!

my personal take is that MT equipment is more or less medium

Interesting. I wonder what you use.

Considering your replies and my situation:

  1. How do I choose the right antenna model from that catalogue?
  2. How do I choose the proper router/modem to use with the proper antenna(e)? (I suppose I will benefit from the cell locking functionality, so I hope you can recommend considering that too)
  3. Can you recommend anything specific (hopefully with some explanation)?

FWIW, in my discussion with my MNO I was told they had plans to develop their network around but who knows when. Perhaps it is a good idea to have a fairly future proof equipment in case that is possible and not conflicting the current goal.

11

I can’t give you very concrete answers because I don’t know actual conditions at your spot. When we did in-depth trouble shooting gor our customers, we often paid tgem a visit to assess conditions and then made a suggestion.

  1. To choose best antenna from cataligue, one would have to do some diligent measuring. Without it one could only suggest some wide-band antenna with not-so-good performance.

  2. Even harder to tell since there are plenty of modem cards but noone knows if any particular one is supported in ROS. Special features (like channel/cell locking) aren’t supported by all (but I’m sure it’s not necessary most of times and possibly that’s the case in your use case as well …)

  3. Obviously not :wink:

12

When I had those discussions with my MNO they refused to send a technician for a complete test even when explicitly asked many times and even when I offered to pay. All they say is “if you want we can terminate the service” and “we have no technicians for this”. Their competitor does not provide service in this area, satellite Internet is too expensive, so I am pretty stuck and need to do something myself. So:

How (and what) do I measure?

As for modems: you say MT’s are outdated and MT equipment is ~medium. Other technical people say MT is better than the popular brands, so you understand my confusion. I really wonder which equipment is good then (considering the case). For antennae - OK, Iskra, I take your word for it. What about the rest (modem, router)? I wonder how to approach this whole issue.

13

Without a “measurement” phone it’s very hard to do proper measurements. But basically: use directional antenna (preferably a similar to the one you plan to deploy permanently), then measure signal levels of cells present (all frequency bands). “Swipe” antenna around (both horizontally and vertically), mark the directions with best signal levels. Quantities relevant are: RSRP (signal strength) and RSRQ (signal quality) … both are normally negative numbers and higher (i.e. less negative) is better. There’s another quantity often measured (RSSI) which doesn’t tell much sk don’t bother with it.
When measuring, take your time. When modems are not actively communicating with network, they go to “sleep mode” often and measure signal less often, which means that reported values may lag behind … or some signal peak or dip is missed entirely. When actively communicating with network, their measurements are quite affected by orders, received from network, but measurements are made very often (so signal strength reported follows reality better).
Measurement equipment (phones) are running customized firmware which gives much greater control over modem behaviour and vastly increases amount of data reported via diagnostic port of device … and that makes measuring much easier.

When choosing microlocation of antennae, it is important to make sure there are no obstacles near the antenna direction. Fresnel zone, which has to be unobstructed to maximize signal strength, is circular around the signal path (line of sight if it exists, or line of reflected signal otherwise) with diameter increasing towards midway between Tx and Rx antennae. And that includes obstacles like tree tops, they attenuate signal quite well.


Modems that come in MT equipment are older generation and in most cases don’t come close to performance of modern phones (which causes great disapointment of users). This is what I meant by “MT’s modems are outdated”. However, in particular cases they might perform equally well as the more modern ones, e.g. when there’s no CA (which might largely be the case in your case where there aren’t many cells of different frequency bands covering the spot where you want intetnet service).

And by “MT’s equipment is mediocre” I’m talking about MT’s integrated 4G solutions (SXT, ATL) which, as it turns out, have decent performance in higher frequency bands but underferform in low freqency bands … which are often crucial in remote places (most MNOs use low frequency bands to provide maximum coverage and use higher bands to provide capacity in places with higher user density). This characteristics is due to the antenna design (a dish) … where physics can’t be worked around: longer wavelengths (lower frequencies) require physically larger antennae … but most users like to buy small and neat devices.
If one pairs a good (or at least a decent) antenna with decent modem and with a good router, then this can be a winning combination. But it may be bordering to a DIY solution. If we go back to MT portfolio: it should be just fine, if you used a MT router with (as good as possible) LTE modem built in … but connect decent external antenna (or actually a pair of them for 2x2 MIMO). Selection of which of modems available depends on bands “active” at your place, ideally modem should support CA in combinations of any bands present.

14

Don’t disagree with anything @mkx said. I just think OP is overthinking this. The tower load and its backhaul is very likely going to affect the speed even more than gear choices.

I’d offer.. if you’re having these extended convos with your MNO… perhaps asking things that help might help in narrowing your equipment choices. e.g. where are the closest towers located? what bands are support on the tower(s)? will the share the backhaul speeds to base station? Checking cellmapper.net also help to guess at some of this info. It’s this kinda info you’d need get more specific advice.

Also, field test mode on a phone will report some more detail info on the tower and generally what the phone is reading. On iPhone, you use Phone app to call 3001#12345# and you can view stuff like the selected tower and neighbors. On Android, there are some apps that let you read this. This might also give an idea of the signal and bands generally at your location.

15

Generally you’re right. But looking at numbers @OP posted in initial post of this thread my (extensive) expertise says that primary task at hand is to improve signal reception as much as possible. Potential cell tower issues (congestion on LTE interface, backhaul congestion, etc.) can (and will) hinder OP’s experience, but first OP needs at least decent experience (to have something to degrade :wink:).
My expertise also says that in some cases no looking at cellmapper web site (or any other similar web site) will replace proper measurement. No radio propagation models (built into “profesional” tools) are good enough to compute exact signal strength in a random point. Typically they have sigma of 4-7dB and knowing statistics it’s obvious that predicted value can either be precise or can be off by large amount (probability of error of a few sigma is stll higher than a few percent). And since we’re talking about cell border case, error exceeding 10dB can make huge difference (between very decent performance and no service what so ever).

16

Thank you guys.

@Amm0

The conversations I shared are not an ongoing thing. I have been fighting MNO’s terrible “customer support” for 2 months until I got in touch with “technical support” who gave me those answers. The later team also told me that customer support and technical support do NOT share information between each other (some genial company rules, I guess) and the tech support was completely uninformed about all the trouble I am going through. During my last conversation with the tech support I asked explicitly about the geographical coordinates of the cell towers I connect to and they gave them to me but could not tell me which cell ID is for which geo location. Those are not on cellmapper.net but I know the places. I didn’t ask which bands were supported and the other details you mentioned as I wasn’t aware I would need this info.

I don’t have an iPhone. The phones I have in hand run stock Android, LineageOS and Replicant. None of them has a 4G modem. I don’t know if the later is needed to get correct idea of the signal power. Unfortunately, I could not find a working “magic code” (or whatever the proper name is) to check tower info. All I can see is the general info about signal power and ASU.

@mkx

So, to choose the right antenna model from Iskra’s catalog I must measure, and in order to measure I must use a similar antenna, pointing it around until I get best result. Assuming try-before-buy (and test different models for free) is possible, that might work. I have not checked if any distributor would allow that. Sounds unlikely but who knows.

Looking at antenna gain graphics in the brochures of MT’s LHG series I see they are weak in the lower (800 MHz) band but their gain at middle (1.7-2.7 GHz) band is higher (15-18 dBi). OTOH, Iskra’s P-58 L700 UNI (which has a U-MIMO version too) has a fairly uniform gain distribution (10-11 dBi) starting from LTE800 up to LTE 2,6.

Considering my own situation (“Cell 1” being best and working in the 1800 MHz band), would it not be wiser to aim for higher gain in that band? IOW, won’t the 15+ dBi gain of LHG/ATL be better than P-58 L700 UNI U-MIMO’s 11 dBi in the same range?

Or would the extra mid-band gain of (say) LHGG LTE6 kit be compensated negatively by its worse lo-band performance, thus reducing the overall result when CA is (hopefully possible and) active?

What do you say?

17

More saying knowing the bands the carrier offered let you know if a higher-end modem is going to do much for you. And if there was NO low-bands … then the LHG performance in the low-band won’t matter much. :wink:

I don’t have any androids, so can’t say which app, but on Android apps are allowed to read the RSRP SINR etc from the android’s cellular data. There is a list of some here: https://www.wilsonamplifiers.com/blog/best-smartphone-apps-to-find-your-mobile-signal-strength/

Apple does not allow apps access to any cell info (e.g. an app could deduce location without going through privacy controls) – so on iOS there is ONLY the “magic code” for the “field test mode”.

18

The cell with best signal strength is working in band 8 (900MHz) … and it would be a decent choice for serving cell (master) in CA constelation. Indeed available performance is not stellar, B8 is still used for GSM or UMTS by many operators so often there’s only 5MHz available for LTE (B3=1800MHz usually allows for at least 15MHz if not 20MHz).

I can’t decide about best antenna for your particular case as it very much depends on actual radio conditions. Which have to be measured.

A fairly decent application for android is Network Cell Info Lite (there’s also a paid version).

19

Here is my attempt to measure the situation.

Before I explain, I should mention that during the winter the router regularly connected to the good cell tower (Cell 1) and the speed with it was as mentioned in the initial post.

So, the test from today:

Temporarily, I put the router I currently use as high as possible - on the chimney of the house. From there I can see the upper edge of the hill on which I know one of the cell towers is. I can’t be sure if there is direct radio visibility to it as there are some trees obstructing the overall view.

Result:

The router connects to Cell 1 and shows:

CQI: CQI0:12 CQI1:13
RSRQ: -7.0dB
RSRP: -100dBm
RSSI: -73dBm
SINR: 17dB
4G uplink frequency: 17700
4G downlink frequency: 18650
Band: 3

That gave me speeds:

Download: 60-70 Mbps (for a few short times it even touched 77 Mbps)
Upload: 25 Mbps

i.e. the DL is more than x2 faster compared to the best I have ever seen and the UL is x25 faster. Surprisingly, improvement is not only possible but quite significant.

Based on this new info, could you recommend the proper equipment, as discussed?
Also, what additional improvement can I expect from it?

20

I’ll defer to @mkx.

@normis from Mikrotik just did a YouTube video on outdoor LTE, his commentary was:

My thought be since you’re using an omni antenna from current router, the RSRP would likely improve, but your SINR is already pretty good. And since CQI is 12/13 out of 15 already … speed isn’t going to go up by much, at least on the band/tower. Yeah the upload be much higher roof/outside, that’s expected – the tower can push more power to you than any device can push to it. So the upload get attenuated more inside than out.

I’d almost be inclined to say mount an omni outside might be okay IF there were more towers in your area. But my thought be the ATL be worth the slight premium, and a directional antenna have more power in winter which seems like you might need. To @normis point, your current reading are not “really poor”. And that ATL allow more bands in future if the cell network changes… But where best to put it on the roof so it can see as much of the tower as possible is likely more important.