First of all, I cannot make miracles happen.

If you are literally living in the middle of nowhere, with no fibre, no fixed wireless provider, and no usable radio signal from any mobile network, this article cannot manufacture internet out of the air. I am from India, so most of my examples are Indian, but the laws of physics are the same for everybody.

This article is for the person who can catch some 3G, 4G, or 5G, but the signal is weak, unstable, congested, or in the wrong place. Maybe you can get internet on the roof but not inside your room. Maybe one provider works in the morning and dies at night. Maybe you maintain multiple SIM cards because trusting one network is a luxury.

I understand your pain because I have struggled almost all my life to get good internet. At this point, I am one of the most qualified people I know to talk about this particular hustle.

The fibre that reached my road, but never reached me

On 15 August 2019, I watched Prime Minister Narendra Modi give his Independence Day speech on national television. He spoke about the goal of giving every village in India an optical-fibre network and broadband connectivity by the 75th year of Independence. I remember the date clearly. I was 17 years old.

He was not announcing BSNL's Bharat Fiber service. That was a separate product. He was talking about a broader national promise: villages like mine would have fibre and broadband connectivity. A year later, on 15 August 2020, the promise became even more specific: all six lakh (600,000) villages were to be connected within 1,000 days. But the speech that stayed in my head, and the day my waiting began, was 15 August 2019.

To be precise, neither speech promised an FTTH installation in every house. But for somebody watching from one of those villages, the expectation was simple: the backbone would arrive, and a usable last-mile connection would eventually follow. I waited. And waited. And waited. It is now August 2026. I am 24, I still live in the same village, and I have still never had a wired broadband connection at my house.

The government published targets and progress numbers. But announcing them again is not the same as following through. In the only sense that matters to my life - the connection actually reaching me - the government never followed through on the promise. I have used normal wired broadband only once in my life, in a city. That one exception is my entire experience of what most people simply call internet.

Fibre was laid along the road to my village. The panchayat office had connectivity for a brief period around 2020 or 2021, but I never got a connection. In 2022 and 2023, I kept visiting the BSNL office and asking. The answer I repeatedly got was that BBNL was being merged with BSNL and I would have to wait until that process was complete. The merger had indeed been approved by the Cabinet in 2022, but an approval on paper did not bring a cable to my house.

And here is the mother of all ironies: my house is right next to that panchayat office. The fibre reached practically next door. It still never reached me.

Later, from what I could see locally, work under the Jal Jeevan Mission appeared to damage parts of the buried fibre while water pipes were being installed. Road reconstruction and expansion appeared to damage more of it. Infrastructure was put into the ground, different infrastructure came afterwards, and somewhere in between, coordination disappeared.

After years of asking, I stopped pursuing BSNL broadband. Not because the problem was solved by the government, but because private mobile networks finally gave me another jugaad.

That is my grudge with BSNL and the government. Internet is no longer some entertainment product. It is how people study, work, apply for documents, access banking, and build a life. A government that asks rural people to participate in a digital country should first give them a reliable way to enter it.

The government now operates Digital Bharat Nidhi, formerly the Universal Service Obligation Fund. Public money from it is used for mobile and broadband projects in rural, remote, and otherwise underserved places. This does not mean that every rural household is automatically entitled to a permanently free internet plan, but it is money specifically meant to close the connectivity gap. I have not benefited from that either.

I am exactly the kind of person this infrastructure is supposed to help. I am a software developer. I work remotely. I genuinely need the internet to earn a living, yet I have never had reliable access. Never. Not so far. I live beside the panchayat where the fibre arrived, in a country funding rural connectivity projects, and I am still building my own internet out of SIM cards, phones, routers, bamboo, cables, and luck.

My internet history is a chain of jugaads

From around 2017 onward, I tried almost everything that was realistically available to me.

I used multiple JioFi routers, Airtel routers, old phones as hotspots, multiple data plans, and multiple SIM cards. I did not keep extra SIMs for fun. I kept them because one network could disappear at any time.

In 2020, I set up a connection about 100 metres from my house by placing a JioFi or phone hotspot there and buying a TP-Link Wi-Fi extender to drag that signal back home. Sometimes I did the reverse and extended Wi-Fi toward the place where I needed to be. I did not know how to build a proper point-to-point link then. I only knew that one spot had internet and my house did not, so I built a crude chain between them.

Once, I raised a long piece of bamboo above the rooftop and tried to put my phone on top of it, hoping the extra height would improve the signal. Half a brick fell on my hand and injured it badly. That was not the only time this struggle became physical. I have rushed from one place to another just to stop a call from dropping, stood in strange positions for longer than I should have, and had family members get hurt on other occasions while chasing that one patch of usable signal. There is something absurd about being physically injured because you want internet, but that is how badly I needed it.

Then I learned about cellular bands. I used apps to inspect the bands around me, rooted phones when I had to, locked different 4G bands, and moved phones around the house looking for the one place where the signal would hold. Later, I bought a Motorola phone that, with the software it had at the time, let me select bands without rooting it.

During 2022-23, before 5G reached me, I bought the TP-Link Archer MR600 V2 for ₹8,099. I bought it largely for 4G+ and band control. It could lock individual bands and perform carrier aggregation, but its firmware would not let me force the exact band combinations I wanted. In my location, modern phones were often better at catching a weak signal. I even opened the MR600, connected to its board over UART, and installed OpenWrt. It was a painful process, and the 2.4 GHz and 5 GHz Wi-Fi performance became worse on that device. I did a lot of work and did not get the result I wanted.

When Jio 5G arrived, it changed my life for a while. I could lock my phone to n78, the 3,500 MHz band, even with a signal around -123 dBm, connect the phone to my router through USB tethering, and use that phone as the internet gateway for the whole house. That signal reading sounds terrible - and it was - but in my exact location, the connection could still perform better than my 4G alternatives.

Once I had 5G, I never looked back at the old 4G setup. At least, not until recently, when the same MR600 returned as part of my BSNL backup.

In April 2024, I got Jio AirFiber. Even its outdoor unit could not simply sit on my house. I bought a 100-metre Ethernet cable and installed the unit roughly 50 to 100 metres away, held up by bamboo at a spot where signal reception was much better than at my house. It was a more capable version of the same thing I had been doing for years: put the receiver wherever the internet exists, then somehow bring the connection home.

Internet access was relatively good from around April 2024 to April or mid-2025. There were fewer 5G users in my village then. As more people upgraded their phones, the network appeared to become more congested, and the good times slowly ended.

Jio AirFiber was still imperfect: speeds to some websites were slow, Cloudflare WARP sometimes improved the route dramatically, latency was often higher than I wanted, and there were random disconnections. As more connections appeared in my area, performance seemed to become worse.

In February 2026, I switched to Airtel. In my location and during this period, Airtel turned out to be even less reliable. I had a 100 Mbps plan that would sometimes give me 2 Mbps. Even during an off-peak test at 5 AM, it could struggle to reach 30 Mbps. My Airtel SIM in a phone could give me around 100 or even 200 Mbps while the fixed wireless unit in the same general area crawled. Sometimes the connection would stop completely in the middle of my work. Rebooting the provider's router would instantly fix it, but that is not how this should work.

The outages became frequent enough that I started counting them. Eventually, I stopped.

More recently, the MR600 found a second life as my current BSNL backup router. With OpenWrt installed, I use ModemManager and its mmcli command-line tool to control the built-in cellular modem, connect through a BSNL SIM, and keep the modem locked to LTE Band 28 at 700 MHz. Getting BSNL online through this stack was another very painful process. It works sometimes. Sometimes it does not.

For a while, that exact BSNL setup gave me around 30 to 37 Mbps and genuinely saved the day when Airtel failed. Later, the previously strong 700 MHz signal largely disappeared from my location. The MR600 is still locked to that band, but the connection is now intermittent. My best guess is that the nearby tower disabled or reconfigured its 700 MHz service, but BSNL has not confirmed that. I only know that my useful backup stopped being useful. One recharge after another has gone to waste this way.

As of this moment, I maintain three SIM cards for internet. So far, I have had nothing that I can honestly call reliable.

I do not need high speeds. I need stable internet.

This is the biggest thing I have learned.

An unreliable 100 Mbps connection is worse than a stable 30 Mbps connection. I do not care that a speed test can touch the advertised number at 5 AM if Google refuses to load in the middle of my work at 3 PM. I do not care that rebooting the router fixes it. I should not have to restart a locked-down provider device whenever the internet decides to die.

Speed, latency, routing, Wi-Fi quality, congestion, and reliability are different problems. Providers combine them into one shiny number and sell that number to us. Real life does not work like that.

My complaints about Jio, Airtel, and BSNL describe my connection in my village during a particular period. They are not a national ranking of the companies. Your nearest tower, its backhaul, the spectrum deployed there, the number of users sharing it, your device, your house, and even the trees between you and the tower can produce a completely different result.

This is why you should diagnose your own location before buying anything.

Money was not the problem. I had the money, and I spent a lot of it on gadgets, routers, cables, recharges, and multiple SIM cards. If ordinary broadband or a practical satellite option at my address had been available during those years, I could have had a much better experience. Instead, I lost months of my life figuring out how to access the internet. Months of my life. I am not exaggerating.

What you can try

1. Map every signal around your house

Before buying a router, collect SIM cards from every provider that has even a small chance of working. Borrow them if you can. Put each SIM in a reasonably modern phone and walk around your property.

Test beside every window, outside each wall, on the terrace, on the roof, and at any nearby high point you can access safely. Test in the morning, evening, and at night. A tower that looks excellent at 5 AM may be unusable when everybody comes home.

Use an app such as NetMonster to see the network type, band, frequency, and signal readings your phone exposes. You do not need root merely to inspect this information, although Android and phone manufacturers do not expose every field on every device.

Write the results down. For each provider, record:

• The location of the phone.
• Whether it connected to 4G or 5G.
• The band or frequency.
• Signal strength and, when available, signal quality.
• Download, upload, and ping.
• Whether the result remains usable for at least 15 to 30 minutes.

With dBm values, a number closer to zero is stronger: -90 dBm is stronger than -120 dBm. But do not turn one number into a religion. RSRP, SINR, channel width, congestion, and backhaul all matter. My n78 connection at -123 dBm worked for me; that does not make -123 dBm a magic threshold for you.

Also test the mobile connection directly on the phone before blaming the cellular network for a slow computer. If the phone is fast but the computer is slow, your bottleneck may be Wi-Fi, the router, the Ethernet cable, or the computer itself.

2. Understand low bands and high bands

If you live far from a tower, start by looking for spectrum below 1 GHz: 700, 800, 850, or 900 MHz, depending on what your provider has deployed in your telecom circle, meaning your regional licensed service area.

Lower-frequency signals generally travel farther and handle obstacles better. Higher-frequency signals generally have a harder time with distance, trees, and concrete, although the real result also depends on transmit power, antennas, bandwidth, and network design. A low band is not automatically faster or slower; it is usually the place to look when coverage is the main problem.

As of August 2026, Jio lists n28, n78, and n258 as its 5G bands in India. n28 at 700 MHz is the coverage option worth hunting for in remote places. n78 around 3,500 MHz has much more capacity available in many deployments, but it usually needs a closer tower or a cleaner path. If you can receive usable n78, try it before building an elaborate multi-connection system. It may outperform all the jugaads below.

Concrete and trees are the enemy, especially at higher frequencies. They do not create an absolute wall that no signal can cross, but they can weaken a marginal link enough to make it useless. Height and line of sight matter.

3. Use band locking carefully

Band locking forces a phone or modem to remain on selected cellular bands instead of accepting whatever the network chooses. It helped me because my phones sometimes preferred a stronger but congested band, while a weaker band gave me better internet.

Band locking is not always better. It can stop the device from using carrier aggregation, prevent it from moving to a healthier cell, or interfere with calls and normal mobility. First test automatic mode. Lock only after your survey shows that a particular band is consistently better for your stationary setup.

On my Motorola phone, I could select bands without rooting the phone. Do not assume every Motorola, or every regional version and software update, has the same menu. Other devices may expose a service menu, require special tools, require root, or provide no safe method at all. I would not root my main phone merely to experiment with internet unless I understood the security and bricking risks.

There is another term you need to keep separate: carrier aggregation. Carrier aggregation lets a cellular modem use multiple component carriers provided by the same mobile network. It is how 4G+ and some 5G combinations increase capacity. It happens between your modem and the operator. It is not the same as combining a Jio connection with an Airtel connection.

Operator architecture also changes. When I first did these experiments, Jio's standalone 5G network made the controls I wanted practical, while Airtel's non-standalone setup did not. Do not treat that as an eternal rule. Networks, phone firmware, and supported bands keep changing.

4. Put the receiver where the internet is

Sometimes the best internet solution is not a more expensive modem. It is a better location and, within safe USB data and power limits, a longer cable or a properly powered extension.

Find the place where the phone receives the best usable signal. Then either turn on its Wi-Fi hotspot, the simplest option, or connect it by USB to a router that can use the phone as its WAN connection.

My setup was a band-locked phone connected over USB tethering to a router running OpenWrt. The phone received the cellular signal, and the router handled Wi-Fi, Ethernet, DNS, guest networks, and the rest of my home network. Android can also be configured to default to USB tethering when connected, although the exact behaviour varies by phone and screen-lock settings. OpenWrt has an official smartphone USB-tethering guide.

If you want to do this, buy a router with a USB port and confirm that the exact model and hardware revision are supported by the current stable OpenWrt release. Read that model's installation and recovery instructions in OpenWrt's supported-device database before buying or flashing anything. Installing unsupported firmware can brick the router. Do not buy by brand alone. D-Link, TP-Link, and every other company have models that are easy, difficult, or impossible to support. I learned this the hard way.

Be sensible about outdoor placement. Phones hate heat, rain, swollen batteries, unstable power, and lightning. Use shade, weather protection, secure mounting, and a safe power arrangement. Do not balance a phone on makeshift bamboo above a roof, improvise remote power without proper outdoor-rated equipment, or work beneath loose rooftop masonry. I learned that lesson with an injured hand. No signal boost is worth getting hurt.

For Wi-Fi inside the house, 2.4 GHz usually travels farther and handles walls better; 5 GHz usually provides more capacity at shorter range. Channel 1 is not magic. Use the least congested legal channel available in your location. Multiple consumer extenders were the only method I understood in 2020, but they were an inefficient way to cross that 100-metre gap. If I were doing it now, I would investigate a professionally installed outdoor-rated cable or fibre run, or a directional point-to-point wireless bridge with a clear line of sight, instead of chaining extenders or expecting an ordinary indoor access point to cover a kilometre. Follow electrical safety, lightning protection, local spectrum, and transmit-power rules.

5. Be careful when buying a cellular router

Do not trust the words "band locking" or "carrier aggregation" on a product page without checking the exact hardware revision, regional SKU, modem, firmware, supported bands, and combinations.

My Archer MR600 V2 could select bands, but it could not reliably force the carrier-aggregation combinations I bought it for. Version 1 and Version 2 did not offer the same controls. That difference was a deal-breaker for me. The router could still aggregate in combinations it selected itself, so it was not useless; it was simply not the device I thought I was buying.

If a router and its modem are properly supported by OpenWrt, ModemManager is another route. It is a service for controlling many kinds of mobile-broadband modems, including QMI- and MBIM-based devices; mmcli is its command-line client. This can expose useful status and controls and can bring up the cellular data connection. In my MR600 setup, it also gave me the control I needed to keep BSNL on 700 MHz. Exact driver, modem, firmware, and band-selection support varies, so do not assume my commands or results will transfer directly to another router.

A second-hand phone can sometimes be a better cellular receiver than a consumer 4G router while also giving you better diagnostic and band controls. But a permanently tethered phone also has disadvantages: battery wear, heat, occasional USB problems, and the ridiculous fact that you have to dedicate an entire phone to being a modem.

There is no perfect answer here. Buy only after testing the bands available at your house.

6. Treat fixed wireless as a site-specific option

Jio AirFiber, Airtel Xstream AirFiber, and similar fixed wireless services can be much simpler than building a phone-and-router setup. They can also suffer from weak signal, bad placement, tower congestion, provider routing, locked-down hardware, and installation rules.

Ask the provider for an official site visit. A location with a clear path toward the serving tower is more promising, particularly for higher-frequency links. Do not assume that because a phone gets 5G, the provider's fixed wireless unit will be accepted or perform well.

Do not register service at a false address, pay an installer unofficially, or relocate provider-owned equipment away from its approved location. You could violate the service terms, create safety problems, lose support, or become liable for damaged or stolen equipment. Get the installation approved.

7. Keep a backup from a different provider

If internet is essential to your work, one connection in a remote place may not be enough. A cheap secondary SIM from another provider can be worth more than upgrading the primary connection from 30 Mbps to 100 Mbps.

Ideally, the backup should depend on different infrastructure, not merely be another plan served by the same congested tower or affected by the same upstream failure. Keep the backup device charged and test it occasionally. A backup that has silently stopped working is not a backup.

I eventually preferred switching Wi-Fi networks manually over building complicated automatic failover. Manual switching is not elegant, but it has fewer pieces that can break. If you need automatic failover, OpenWrt's multi-WAN tools can do it, but understand that ordinary load balancing does not add the speeds of two connections to a single-flow download.

8. Connection bonding is the final boss

If you have several slow connections and none is good enough alone, you can investigate Multipath TCP and a project such as OpenMPTCProuter (OMR). This is different from cellular carrier aggregation. It uses multiple internet paths and a remote server to aggregate traffic or fail over between them. The Linux kernel documentation explains the underlying idea.

For example, three independent connections delivering about 4 Mbps each might approach 12 Mbps together under favourable conditions. You will not always get the arithmetic sum. Protocol overhead, latency differences, packet loss, the remote server, and the destination can all become bottlenecks.

I tried OpenMPTCProuter. The setup worked, but it was not worth it for me. I needed multiple phones or modems, several internet plans, a compatible router, and a VPS. Six devices ended up involved. Some social media, government, banking, and shopping websites I visited treated the VPS address as suspicious or blocked it. The setup had more latency, more cost, more points of failure, and no useful IPv6 path in my configuration. Sometimes OpenMPTCProuter itself went down. Under load, it could perform worse.

Most importantly, I was combining all that machinery to get around 30 or 40 Mbps when one direct connection could already give me roughly 30 Mbps. The marginal utility was almost zero.

If you have four 8 Mbps lines and combining them is the only realistic way to get a usable connection, OMR could be life-changing. If you already have a mostly stable 30 Mbps line, think very carefully before constructing the final boss.

The order I would try today

If I had to start again, I would keep the order simple:

1. Check every official wired, fixed wireless, satellite, and BharatNet option first.
2. Survey every mobile provider around the house before buying hardware.
3. Put the best SIM in a compatible phone and try a hotspot from the best location.
4. Add an OpenWrt router and USB tethering only if the whole house needs that connection.
5. Keep a modest backup from a different provider.
6. Leave point-to-point links, multi-WAN, and OpenMPTCProuter until the simple options have genuinely failed.

Do not make an internet decision in haste. I repeatedly responded to one bad week by buying another device, changing another plan, or constructing another fragile setup. Sometimes the new thing was better. Sometimes it only gave me a new category of problem.

What I have finally accepted

I do not need 100 Mbps. I do not need a speed-test screenshot. I need the internet to be there when I click something in the middle of my work.

Stable 30 Mbps is enough for almost everything I do. Unstable 100 Mbps is not. BSNL can save the day and then disappear. Jio can be the best option available and still have bad routes or congestion. Airtel can work beautifully on a phone and fail through its fixed wireless hardware. There is no loyalty or ego left in this for me. Credit where it is due, criticism where it is due, and always keep a backup.

I have fought this battle with JioFi boxes, Airtel routers, old phones, rooted phones, Motorola service menus, USB tethering, OpenWrt, an Archer MR600, Jio AirFiber, Airtel AirFiber, BSNL 4G, three SIM cards, and OpenMPTCProuter.

Even now, basic reliable connectivity still feels like a luxury.

People who need it the most do not get it.

My work depends on the internet.

If my country requires me to leave my house and move to a city merely to get a good internet connection, I would much prefer to move to a city called San Francisco instead.

One day, I will have a one-gigabit-per-second internet connection. I know it will happen. I am eagerly waiting for that day.

None of this would have happened if I had basic broadband connectivity. None of it. I did not choose this fight or build all these jugaads because I enjoyed them. I had to fight my way out because I had no other choice. I still don't.

Every moment of stable, reliable internet is a blessing.