Introduction to LPWAN and the Role of Phase Modulation

Low Power Wide Area Networks (LPWAN) have emerged as a cornerstone of the Internet of Things (IoT), enabling connectivity for devices that require long-range communication, low data throughput, and years of battery life. Examples include smart meters, soil moisture sensors, asset trackers, and environmental monitors. The physical layer (PHY) of any LPWAN technology is where modulation choices directly impact power efficiency, range, and robustness. Among the various modulation families, phase modulation plays a central role because it offers excellent noise immunity while keeping energy per bit low. This article examines the phase modulation techniques used in LPWAN, explains how they work, and provides a detailed comparison of their performance in real-world IoT deployments.

Basics of Phase Modulation

Phase Shift Keying (PSK) Fundamentals

Phase modulation encodes digital information by shifting the phase of a sinusoidal carrier signal. In its simplest form, the carrier phase takes one of a finite set of discrete states, each representing a group of bits. A phase shift of 0° might represent a binary 0, while a shift of 180° represents a binary 1 in Binary Phase Shift Keying (BPSK). Because the information is carried in the phase rather than the amplitude, PSK schemes are less vulnerable to amplitude noise and fading compared to amplitude-based methods. This property makes PSK particularly attractive for LPWAN links where signals must survive long propagation paths with low transmit power.

Constellation Diagrams and Symbol Mapping

The set of allowed phase states is visualised in a constellation diagram, where each point corresponds to a specific phase (and often amplitude) of the carrier. For pure phase modulation, all points lie on a circle of constant amplitude. BPSK has two points at 0° and 180°; QPSK has four points at 45°, 135°, 225°, and 315°. The distance between adjacent constellation points determines the noise margin: larger distances yield lower bit error rates (BER) at a given signal-to-noise ratio (SNR). In LPWAN systems, where the SNR is often near the receiver sensitivity limit, robust constellations with good phase separation are desirable.

Key Phase Modulation Techniques in LPWAN

Binary Phase Shift Keying (BPSK)

BPSK is the most robust phase modulation scheme, offering a symbol error probability that is about 3 dB better than QPSK for the same energy per bit. It encodes one symbol per bit, making it spectrally inefficient but highly resilient. BPSK is used in several LPWAN standards for the control channels or for the payload when maximum range is required. In NB-IoT, for example, the narrowband physical downlink control channel (NPDCCH) employs BPSK to ensure reliable decoding even at the cell edge. The trade-off is low data rate: at a given bandwidth, BPSK achieves only half the throughput of QPSK. However, for applications that only need to transmit a few bytes per day, BPSK's simplicity and power efficiency often outweigh the lower data rate.

Quadrature Phase Shift Keying (QPSK)

QPSK doubles the data rate of BPSK without increasing the required bandwidth by encoding two bits per symbol via four phase states. The four constellation points are typically rotated 45° relative to the axes to simplify carrier recovery. QPSK is widely used in LPWAN technologies that need higher throughput, such as NB-IoT for downlink data transmissions. The symbol error rate for QPSK in an additive white Gaussian noise (AWGN) channel is the same as for BPSK when energy per symbol is constant, but because each symbol carries two bits, the bit error rate is slightly worse under the same energy per bit. Nonetheless, the bandwidth efficiency of 2 bits per symbol per second per hertz makes QPSK a practical choice for many LPWAN use cases.

Offset QPSK (OQPSK)

A variant of QPSK, Offset QPSK delays the quadrature (Q) component by half a symbol period relative to the in-phase (I) component. This simple modification prevents simultaneous 180° phase transitions, which cause large instantaneous amplitude variations that can distort the signal when using non-linear power amplifiers. OQPSK is especially useful in LPWAN endpoints that employ highly efficient but non-linear Class C or Class E amplifiers to maximise battery life. By limiting phase jumps to 90°, the envelope remains nearly constant, reducing spectral regrowth and interference to adjacent channels. OQPSK is used in the IEEE 802.15.4 standard, which underpins several LPWAN protocols, and also appears in some proprietary LPWAN implementations.

Differential PSK (DPSK) and pi/4-DQPSK

In differential phase shift keying (DPSK), the information is encoded in the phase change between consecutive symbols rather than in the absolute phase. This eliminates the need for a coherent carrier reference at the receiver, simplifying the demodulator and reducing power consumption. DPSK is widely adopted in LPWAN systems where achieving phase coherence is costly or impractical. The pi/4-DQPSK variant rotates the constellation by 45° on every symbol, limiting phase changes to ±45° and ±135°. This form is used in some narrowband IoT standards because it combines differential decoding with reduced envelope variation, making it robust against fading and frequency offsets. While DPSK suffers a penalty of about 2-3 dB in SNR compared to coherent PSK, the savings in receiver complexity and the ability to operate without a phase-locked loop often make it the better choice for low-cost, low-power IoT sensors.

Comparison of Modulation Schemes for LPWAN

Selecting the appropriate phase modulation scheme for a LPWAN system involves balancing data rate, range, power consumption, and receiver complexity. BPSK offers the best sensitivity but the lowest spectral efficiency. QPSK improves throughput at the cost of 3 dB more energy per bit for the same BER. OQPSK and pi/4-DQPSK trade a small degradation in BER (compared to QPSK) for better performance under non-linear amplification and frequency offset. For battery-powered devices that transmit infrequently, BPSK is often favoured for the uplink when range is critical, while QPSK or DPSK may be used for the downlink where the base station has more power and can afford a coherent receiver. In hybrid LPWAN technologies like LoRa, the modulation is actually chirp spread spectrum (CSS) which is not purely phase-based but incorporates continuous phase changes; however, many commercial LoRa implementations also support FSK with PSK-like properties for back-channels.

Phase Modulation in Major LPWAN Technologies

LoRa and Chirp Spread Spectrum

LoRa (Long Range) uses a proprietary chirp spread spectrum (CSS) technique where the carrier frequency is swept linearly over time. While CSS is fundamentally a frequency modulation, the chirps can be viewed as a form of continuous-phase modulation. A chirp that starts at a certain frequency and increases or decreases encodes information by the timing of its frequency zero crossings, which is essentially a phase-modulated signal. LoRa receivers are designed to be highly sensitive (down to -148 dBm) thanks to the processing gain of the spread spectrum. The modulation supports spreading factors from 7 to 12, trading data rate for range. For the LoRaWAN standard, the modulation remains CSS on the uplink, but the downlink often uses Gaussian Minimum Shift Keying (GMSK) or continuous-phase frequency shift keying (CPFSK) which are related to phase modulation through the phase continuity property.

NB-IoT and LTE-M

NB-IoT, part of the 3GPP Release 13, uses a narrower bandwidth (200 kHz) with subcarrier spacing of 3.75 kHz or 15 kHz. The downlink employs Orthogonal Frequency Division Multiple Access (OFDMA) with QPSK modulation on each subcarrier. The uplink uses Single-Carrier Frequency Division Multiple Access (SC-FDMA) with π/2-BPSK or π/2-QPSK. The π/2 prefix indicates a half-rotation per symbol, which reduces the peak-to-average power ratio (PAPR) and improves power amplifier efficiency – a critical factor for battery-operated IoT devices. LTE-M (Cat-M1) uses a wider bandwidth (1.4 MHz) and also relies on QPSK (and 16QAM for high-capacity scenarios) but with similar PAPR-reduction techniques. Both NB-IoT and LTE-M benefit from the inherent noise immunity of PSK schemes while achieving data rates up to 250 kbps and 1 Mbps respectively.

Sigfox and Ultra-Narrowband

Sigfox uses an ultra-narrowband (UNB) approach with a bandwidth of only 100 Hz per channel. The modulation is Differential Binary Phase Shift Keying (DBPSK) on the uplink and Gaussian Frequency Shift Keying (GFSK) on the downlink. DBPSK is a natural fit for UNB because it can be demodulated non-coherently, lowering cost and power. The very narrow bandwidth allows the receiver to achieve excellent sensitivity (down to -142 dBm) while keeping the transmit power low (typically 25 mW). Sigfox’s use of DBPSK, combined with a low duty cycle and a random frequency-hopping pattern, provides robustness against interference and fading. The trade-off is a very low data rate of 100 bps (uplink) and 600 bps (downlink), which is sufficient for applications like temperature monitoring or leak detection.

Trade-offs: Power, Data Rate, and Range

No single phase modulation scheme is optimal for all LPWAN scenarios. BPSK and DBPSK offer the longest reach at the lowest data rate, making them ideal for battery-powered sensors that send a few bytes per day from basements or rural areas. QPSK and OQPSK double or triple the data rate but reduce the link budget by about 3 dB – meaning the range shrinks by roughly 10-20% for the same transmit power. In practice, LPWAN designers often use a combination: the uplink (sensor to gateway) uses robust BPSK or DBPSK to maximise battery life, while the downlink (gateway to sensor) may use QPSK because the gateway has ample power and can afford a more complex receiver. The choice also depends on the allowed duty cycle and regulatory constraints; for example, in the EU 868 MHz band, devices must limit transmit duty cycles, favouring schemes that minimise time-on-air (i.e., higher data rate).

Implementation Considerations

Synchronization and Phase Noise

All PSK systems require accurate frequency and phase synchronisation between transmitter and receiver. Low-cost IoT crystal oscillators have limited accuracy (typically ±20 ppm), leading to frequency offsets that can rotate the constellation and cause errors. Differential schemes like DPSK inherently tolerate small offsets because they decode based on phase changes rather than absolute phases, but they still require symbol timing recovery. OQPSK and π/4-DQPSK are more tolerant of phase noise than plain QPSK because they avoid large phase jumps. In LPWAN receivers, automatic frequency control (AFC) and phase-locked loops (PLL) are often omitted to save power; instead, non-coherent or differentially coherent detection is used. For example, LoRa receivers employ a timing loop that tracks the chirp starting point, effectively implementing a form of delay-locked loop.

Transceiver Design for Low Power

The power amplifier (PA) is one of the biggest consumers of energy in a wireless transmitter. PSK modulations that maintain a constant envelope, such as BPSK and QPSK (when using a linear PA), allow the PA to operate near saturation where efficiency reaches 50-60%. However, QPSK with a raised-cosine pulse shaping introduces amplitude variations that reduce efficiency. OQPSK and π/4-DQPSK, by limiting phase transitions, keep the envelope nearly constant and are thus favoured in highly efficient PAs. More advanced techniques include polar modulation, where the phase and amplitude components are generated separately and combined at the PA – this allows very high efficiency but adds complexity. Many LPWAN chipsets integrate a digitally controlled oscillator (DCO) and a phase modulator to generate the PSK signal directly in the digital domain, reducing analog component count and power.

Research continues into advanced phase modulation for LPWAN, aiming to push the boundaries of sensitivity and spectral efficiency. One promising direction is the use of continuous phase modulation (CPM), of which GMSK and GFSK are special cases. CPM signals have a constant envelope and smooth phase transitions, offering excellent power efficiency while achieving spectral compactness. Another trend is the integration of frequency-hopping spread spectrum (FHSS) with PSK to combat interference, as seen in some proprietary LPWAN systems. Additionally, the emergence of massive MIMO (multiple-input multiple-output) in cellular LPWAN (e.g., NB-IoT in 5G networks) may allow coherent detection of multiple narrowband PSK streams, improving capacity. On the receiver side, machine learning-based detectors that learn the phase error characteristics of a given hardware platform are being developed to further reduce the required SNR. These advances suggest that phase modulation will remain a key enabling technology for LPWAN as the IoT expands to billions of devices.

Conclusion

Phase modulation techniques underpin the physical layer of many prominent LPWAN technologies. From the robust simplicity of BPSK in NB-IoT control channels to the differential DBPSK used in Sigfox ultra-narrowband, each scheme offers a unique balance of range, data rate, and power consumption. Understanding these trade-offs is essential for IoT system architects who need to select the right modem or protocol for a given application. As LPWAN evolves to support higher device densities and more demanding use cases, phase modulation will continue to be refined, driven by the constraints of low energy and long range. For engineers deploying LPWAN solutions, a solid grasp of phase modulation fundamentals enables better design decisions and more reliable field performance.

For further reading, consult the Semtech LoRa modulation white paper, an IEEE survey on LPWAN modulation schemes, and the Wikipedia article on PSK fundamentals. Coverage of NB-IoT modulation can be found in 3GPP technical reports on NB-IoT, and more details on Sigfox modulation are available in Sigfox's technology overview.