5G

5G Deployment and Troubleshooting: What Actually Works

Most people talk about 5G like it is a single technology. It is not. It is three separate things running on top of each other, and treating them as identical will get you into trouble if you are setting this up for anything beyond basic mobile browsing. The sub-6 spectrum, the mmWave bands, and the mid-band spectrum each behave completely differently in the field. I have spent the last several years dealing with carrier rollouts and enterprise deployments, and the number of projects that stall because someone assumed mid-band would cover what they needed is genuinely high. The critical distinction most guides skip is non-standalone versus standalone architecture. NSA is what you will encounter in the vast majority of consumer deployments right now. It uses the existing 4G LTE core for control signaling and only offloads data to the 5G New Radio layer. SA drops that requirement entirely and runs on a pure 5G core. The practical difference matters because latency numbers you see advertised assume SA. On NSA, your real-world latency will often sit in the 20 to 40 millisecond range depending on how congested the LTE anchor cell is, not the single-digit figures carriers push in marketing material. Mid-band 5G typically occupies the 2.5 to 3.7 GHz range, sometimes extending to 4.2 GHz in certain markets. This is the band that actually provides useful coverage radius while delivering meaningful speed improvements over LTE. A typical mid-band cell on 100 MHz of bandwidth with 256-QAM modulation can sustain around 800 megabits per second to a single device under clean conditions. The coverage footprint is roughly equivalent to a well-placed LTE cell, maybe 30 to 50 percent smaller depending on terrain and building materials. This is where most of your real deployment work happens.

Millimeter wave operates above 24 GHz, usually between 24 and 39 GHz in current deployments. The physics here are brutal. Rain attenuation becomes a real factor at these frequencies. A light drizzle can introduce 3 to 5 dB of additional path loss. Dense foliage will eat through the signal within a few meters. The penetration loss through a standard concrete wall at 28 GHz is typically around 20 to 30 dB, which means you will lose indoor connectivity almost entirely unless you have a dedicated indoor small cell. The upside is raw throughput potential. With 400 MHz of contiguous spectrum and massive MIMO arrays, single-user speeds above 2 gigabits per second are achievable in controlled environments. I ran into a specific issue last year at a manufacturing facility where we were deploying 5G private network infrastructure. The site had extensive steel framing and concrete floors that killed the mid-band signal before it could reach the production floor. We initially tried adding more macro-level cells, which just created co-channel interference without solving the coverage gap. The workaround was installing low-power distributed antenna systems on each floor with localized small cells tuned specifically for the ASR-2600 spectrum requirements rather than pushing more power through walls that would not let it through anyway. That cut our dead zone time from several hours of troubleshooting down to about 45 minutes of hardware repositioning.

Planning a Realistic Deployment

Site selection is where most 5G projects go sideways. You need to account for line of sight on mmWave links, which means every corner, overhang, and piece of outdoor equipment becomes a potential signal blocker. For mid-band, you can tolerate some NLOS conditions but you should still plan for it. A rule of thumb that saves a lot of rework: calculate your cell spacing at roughly 70 percent of what the theoretical propagation models predict. You will thank yourself later when actual building materials and environmental factors show up. Spectrum licensing varies by country and often by region within countries. In the United States, the C-band auction allocated 39 GHz of spectrum in the 3.7 to 4.2 GHz range, while Europe largely relied on the 700 MHz and 3.4 to 3.8 GHz bands. If you are planning an enterprise deployment, check what local regulators have made available before you buy any hardware. Buying equipment tuned to frequencies that are not licensed in your area is a waste of budget and time. Backhaul capacity is another area where people underestimate requirements. A single 5G cell on mid-band can easily consume 1 to 2 Gbps during peak usage in a dense urban environment. If your backhaul link is anything less than 10 Gbps fiber with headroom, you are going to bottleneck the cell before it ever reaches its theoretical maximum. Wireless backhaul solutions exist but they introduce their own latency and capacity constraints that compound quickly under load.

Testing and Validation

Field testing with proper equipment is essential. A smartphone speed test app will give you a rough idea of coverage but it tells you nothing about signal quality, interference patterns, or actual throughput under load. Tools like Q-Cell or TEPSA with a 5G-capable test modem will give you RSRP, RSRQ, and SINR measurements across the entire bandwidth you are deploying. RSRP below minus 100 dBm usually means you are in dead zone territory on mid-band, and SINR below 5 dB indicates interference problems that no amount of power adjustment will fix. Throughput testing should be done at multiple times of day across different days. Network congestion varies dramatically between 9 AM on a Tuesday and 2 PM on a Friday. I typically run tests during off-peak hours first to establish baseline performance, then during peak hours to see how the network degrades. The difference between those two measurements is what your users will actually experience during normal operation.

Common Pitfalls to Avoid

Handover failures between 4G and 5G cells are one of the most persistent issues in NSA deployments. If your LTE anchor cell and 5G NR cell are not properly aligned in terms of timing and parameter configuration, devices will drop connections during handover instead of transitioning smoothly. This usually manifests as video calls freezing or large file transfers failing at random intervals. The fix involves checking your X2 interface configuration between the eNodeB and gNodeB and ensuring your measurement reporting thresholds are set appropriately for the specific frequency bands you are using. Another frequent mistake is assuming that more spectrum always equals better performance. Spectrum aggregation helps, but only when the component carriers are properly synchronized and the device supports the aggregation configuration. I have seen deployments where carriers aggregated three separate frequency blocks but devices were only connecting to one because the aggregation parameters were misconfigured in the RRC connection setup. The result was worse performance than a single clean carrier because the device was spending processing overhead on a configuration it could not actually use. Power consumption is a limitation worth noting. 5G equipment, particularly mmWave small cells with active antenna arrays, draws significantly more power than equivalent LTE gear. A typical mmWave small cell can consume 150 to 250 watts under full load compared to 50 to 80 watts for a comparable LTE unit. If you are deploying in an existing facility without adequate power infrastructure, you may need to upgrade circuits or install localized power solutions before equipment arrival.

The technology has real limitations beyond coverage and cost. Battery drain on subscriber devices remains a genuine issue, though this has improved noticeably since the initial 5G phone wave. Enterprise customers should also be aware that 5G network slicing, while technically feasible, requires significant configuration effort and is only truly effective when you have dedicated spectrum and core infrastructure. Shared-slice environments across multiple tenants introduce complexity that often outweighs the benefits unless you have very specific isolation requirements.