AI devices are getting smarter and more connected. But when multiple radios—WiFi, Bluetooth, Cellular, GNSS—work inside the same box, interference becomes a real headache. This post shares a real case study from our design team and how we solved the coexistence problem.
Why Coexistence Matters in AI Device antenna
AI edge devices—whether smart cameras, autonomous vehicles, or industrial control units—often operate in harsh RF environments. Their onboard radios share limited physical space and frequency spectrum. Poorly planned systems may experience:
- Degraded GNSS accuracy due to LTE harmonics
- Wi‑Fi packet loss when operating near Bluetooth
- Thermal issues from tightly packed transceivers
- Reduced throughput due to internal EMI or antenna detuning
- Proper coexistence planning is the key to overcoming these challenges.
Antenna Design: The First Line of Defense

We recently worked with a customer on a new AI device. Small box, big ambition. Inside: Wi-Fi, Bluetooth, Cellular, GNSS. All four radios run at the same time.
we expected a straightforward task—pick the right antennas, match them with cables, and send test samples. But as soon as all four radios—WiFi, Bluetooth, Cellular, and GNSS—were powered up inside the same enclosure, a serious challenge emerged: crippling interference across nearly every signal,GNSS accuracy dropped. Bluetooth disconnected. Wi-Fi slowed down. LTE was too strong.
At NextG, we specialize in industrial and embedded antennas. We don’t design the whole product. Instead, we partner with engineering teams during the RF validation phase—after initial layout and antenna selection—to review performance, identify coexistence issues, and help optimize antenna systems for reliable multi-radio operation.
Here’s what we found. And how we fixed it.
Whether you’re prototyping an IoT gateway, smart device, or AI-powered edge system, wish these real-world insights would be helpful for you.
Problem 1: GNSS Accuracy Collapsed in the AI Device antenna
From the test logs, GNSS accuracy was off by hundreds of meters whenever the LTE radio transmitted. They had selected a high-gain active GNSS antenna from our catalog, which looked fine during standalone testing—but real-life isn’t isolation testing.
We quickly suspected LTE Band, which overlaps harmonically with the GNSS L1 band (1.575 GHz)—a common culprit in multi-radio antenna design.
- Root cause: The 4G transmission leaked into the GNSS band, saturating the receiver.
- Effect: GNSS fix rate went down, and location accuracy failed to meet industrial standards
We suggest:
- Adjusted antenna placement to create better isolation.
- Added a filter to block out-of-band noise.
- Tuned the matching circuit to maximize GNSS sensitivity.
Result: GNSS fix recovered, accuracy stabilized.
Tips: interference often comes from neighbors, not the antenna itself.
Problem 2: WiFi vs Bluetooth (2.4 GHz Band Coexistence of the AI Device antenna)
The device used BLE to connect to sensors and 2.4G/5G Wi‑Fi for cloud data. Every time Wi‑Fi uploaded large packets, Bluetooth connections dropped.
Looking at their layout, both antennas were placed on the same PCB edge, with shared ground and nearly identical orientation. This is a classic mistake in design for AI device antenna.
We suggest:
- Designed a shared antenna with proper impedance matching.
- Applied time-domain coexistence algorithms so WiFi and Bluetooth took turns instead of colliding.
- Verified with OTA (over-the-air) tests to confirm stability.
Result: Smooth WiFi streaming and stable Bluetooth audio at the same time.
Tips: in antenna design, space and orientation beat software tricks.
Problem 3: LTE Too Loud
LTE uplink was drowning everything else. Wi-Fi and GNSS SNR both dropped.
After checking, we found it caused by the 5 dBi whip antenna in the device. Normally people think “high gain is better”, but in the small device, that was overkill.
We suggest:
- Switched to a moderate-gain adhesive antenna with better pattern control.
- Helped tune the matching circuit for real-case SWR.
- Added shielding tape around the LTE cable inlet.
After this, signal quality across all radios noticeably improved. Result: clean coexistence across radios.
Tips: more gain ≠ better. Overpowered antennas create more problems than they solve.
The testing. Real Coexistence Testing = Final Enclosure + Final Cables
One last thing we always insist: test antennas in their final mechanical configuration.
Our customer initially ran RF testing with bare PCB and open-frame antennas. Once they closed the housing, SWR shifted, and isolation dropped.
We suggest:
We provided custom-length RF cables to avoid folding near antennas.
Switched one antenna to an FPC layout, which gave them more routing freedom.
Measured isolation and retuned antenna positions using real-time VNA + spectrum analyzer in their product housing.
How We Approached Multi-Radio Coexistence
From this case, a few lessons stand out:
- Always check harmonic overlap between radios.
- Use RF filters to separate sensitive bands like GNSS.
- Plan antenna isolation early—PCB layout and enclosure matter a lot.
- Validate with real-world OTA testing, not just simulation.
Key Takeaways for Engineers
- Multi-radio coexistence is not just theory; it impacts real products.
- GNSS is highly sensitive to leakage from Cellular bands.
- WiFi and Bluetooth must be coordinated to share the 2.4 GHz band.
- Proper antenna design, filtering, and coexistence algorithms are essential for AI edge devices.
Want to Learn More?
If you’re designing AI or IoT devices with multiple antennas, check out:
- AI & IoT Antenna Solutions
- Related article: How Many Antennas Does an AI Box Need?