Definition
Modulation is the process of mapping baseband digital signals onto RF carriers (amplitude, phase, frequency) for wireless transmission; Demodulation is the inverse process at the receiver to recover the baseband signal from the carrier. The modulation scheme directly determines spectral efficiency (bps/Hz), noise resistance, and multipath resistance. LTE/5G NR air interface mainly uses high-order modulations such as QPSK, 16QAM, 64QAM, 256QAM, combined with channel coding and OFDM to achieve high throughput and reliable transmission.
Modulation Methods and Characteristics
- QPSK (Quadrature Phase Shift Keying): 2 bits/symbol, 4 phase points, the most robust, used for control channels, weak coverage scenarios, and 5G NR PBCH and PDCCH
- 16QAM/64QAM/256QAM: 4/6/8 bits/symbol, higher-order modulation improves spectral efficiency, requires higher signal-to-noise ratio (SNR), used for high-speed transmission on data channels (PDSCH/PUSCH)
- π/2-BPSK: adopted in 5G NR uplink, reduces peak-to-average power ratio (PAPR), beneficial for terminal power amplifier efficiency and uplink coverage
- 1024QAM: introduced in 5G-Advanced, used for ultra-high throughput under excellent channel conditions
Adaptive Modulation and Coding (AMC)
The system dynamically selects modulation order and coding rate (MCS) based on CQI (Channel Quality Indicator) reported by the terminal, achieving a trade-off between rate and reliability: high-order modulation is used under good channel conditions to boost throughput, while lower-order modulation is used under poor channel conditions to ensure block error rate (BLER). AMC, combined with power control and MIMO beamforming, is one of the core mechanisms for enhancing spectral efficiency in LTE/5G.
Typical Use Cases
Modulation and demodulation technology runs throughout the entire cellular network link: downlink broadcast channels use QPSK to ensure coverage, data channels adopt 16QAM–256QAM based on scheduling; uplink can use 64QAM for close points, QPSK for distant or edge points. IoT terminals in weak coverage scenarios are generally assigned QPSK or lower-order modulation by the base station, sacrificing speed for connection stability.
When SimtoLink's IoT devices transmit data over cellular networks, both their uplink and downlink rely on modulation and demodulation technology. In weak coverage areas such as industrial sites and underground garages, devices may operate for extended periods using low-order modulation schemes like QPSK, where the rate is limited but the connection is stable. When selecting modules for customers and evaluating network quality, SimtoLink pays attention to the devices' actual modulation order and CQI feedback. Combined with coverage optimization recommendations, this helps customers achieve the best balance between speed and reliability.
Constellation Spacing Sets the SNR That Is Required
Modulation maps a group of bits onto one combination of carrier amplitude and phase. QPSK has only four points. The distance between them is large, so a little noise that twists the constellation can still be decided correctly. Control channels and the coverage edge almost always use it. 16QAM, 64QAM, and 256QAM pack more points into the same power range. Each symbol carries more bits, but the same phase error steps into a neighboring point and the bit error rate rises. Demodulation decides the received point back to the nearest legal combination by a threshold. A wrong decision does not show up in the application as an occasional checksum failure. It first becomes a block error at the physical layer.
Coding and modulation are chosen together. The same modulation order can be paired with different code rates. When the channel is only slightly worse, 16QAM can stay and the redundancy can increase. There is no need to fall straight back to QPSK. Talking only about “it is 64QAM right now,” without the code rate, says nothing about robustness at that moment. Uplink also has a peak-to-average problem. High-order modulation stacked with multiple carriers forces the power amplifier to back off. An edge device that wants a high order may still be unable to transmit it.
Raising and Lowering the Order Needs Stable Quality Feedback
Adaptive modulation and coding works only when the channel quality the device reports is timely and the base-station outer loop corrects toward the target block error rate. Raise the order when the channel is good and the feedback can be trusted. Lower it when the block error rate exceeds the target. At high speed, in a deep fade, or when feedback is switched off, stay on a low order. Otherwise every transport block is retransmitted. If an IoT packet is only a few dozen bytes, the airtime saved by climbing to 256QAM is very limited, and it demands near-site SNR that most metering and sensing cases do not have.
Also confirm that the device category really supports the target order. If the module only implements up to 64QAM, turning 256QAM on in the cell gives it nothing. The scheduler still stops inside the device’s capability. Locking a high order by hand under weak coverage makes the traffic channel fail continuously after random access. It looks like a successful attach followed immediately by a drop. Let rate control move with quality. Do not write a fixed MCS into the module.
Writing the Peak Order Down as the Rate for the Whole Link
The peak rate in a brochure usually assumes the near site, high-order modulation, multiple antennas, and a fairly wide bandwidth all at once. If the field stays on QPSK for a long time, the rate can be an order of magnitude lower. That is the link budget, not a fault. Another mistake is to see a high order only during a download speed test and assume the uplink is the same. Uplink power and peak-to-average ratio are tighter. Many IoT devices stay on QPSK or 16QAM in the uplink for a long time. Accepting an uplink video stream or a bulk report with a downlink speed test only shows the gap after go-live.
An outer-loop target set too strictly also causes trouble. Pressing modulation to the bottom in pursuit of zero block errors lowers cell throughput. Edge users occupy the air longer, which then affects everyone else. Watch block error rate, the modulation distribution, and retransmission count together. A high block error rate that stays on a high order means feedback or the outer loop has failed. A very low block error rate that is forever QPSK means a lock, or a link budget that was underestimated.
The outer loop needs a number of transport blocks before it can move the target SNR to a new point. If the service sends only one small packet an hour, the outer loop has almost no samples. Modulation stays at the initial value. That can be too conservative, or the first choice can be too high. Give the initial modulation an explicit conservative default. Do not expect adaptation to converge inside a single packet. When the code rate changes, the same constellation points do not have the same resistance to interference. Recording only the modulation order treats a looser code rate as a better channel.