Mining over LTE and mobile internet

Is mobile internet good enough for mining?

Yes, if the line is stable. Mining needs very little bandwidth, so the speed of an LTE plan is rarely the problem. What matters is latency, the time a message takes to reach the pool and come back, and stability.

A rig exchanges a few kilobytes per share with the pool. Even a modest LTE plan carries the stratum traffic of hundreds of rigs. Stratum is the protocol your miner uses to talk to the pool.

Every extra millisecond delays job delivery and share submission. Every reconnect throws away work in flight. LTE's typical 30 to 80 ms is workable, but its variability is the real enemy.

How do you set up an LTE site for stability?

• Use a proper LTE router with external antennas, not a phone hotspot.
• Aim the antennas, and lock the router to the band that tests best.
• Check whether the plan gives a real routable connection or carrier-grade NAT (CGNAT), where one public address is shared by many customers.
• Mining works behind CGNAT. Remote access to your rigs will need a VPN or a tunnel.
• Watch data caps. A rig's stratum traffic is small, but OS updates, dashboards and remote desktops are not.

Which server region should an LTE rig use?

Pick the region by measurement, not by the map. Mobile carriers route traffic through their own gateways. So the region closest on the map is not always the closest in milliseconds.

• Run the latency test on the How to Mine page from the LTE connection itself.
• Pick the region it recommends.
• Test again if you move the antenna or the carrier changes its routing.

How do you keep dropouts cheap?

LTE lines drop for a few seconds now and then. Modern miners reconnect and resume on their own. Your job is to keep each interruption cheap:

• A watchdog, a tool that restarts the miner after a rare hang.
• Worker-offline notifications, which wait out a short gap so a routine blip does not raise one.
• A weekly look at your stale-share percentage.

A stale share is work that reached the pool after the job had changed. If stale shares creep up, re-run the latency test before you blame the rig.