Offset-QAM-16 Modulation Overview

• Offset-QAM-16 is a modulation scheme that shifts the conventional 16-QAM constellation to improve phase recovery and receiver performance.
• It employs silicon photonic transmitter architectures, achieving a normalized amplitude contrast of approximately 0.64 and a pre-FEC BER of 10⁻⁶ at 400 Gb/s.
• In multicarrier systems like OFDM/OQAM, Offset-QAM-16 enhances spectral efficiency by eliminating cyclic prefixes and providing SIR improvements between 40–60 dB.

Offset-QAM-16 is a modulation format characterized by the application of a controlled offset to the classical quadrature amplitude modulation (QAM) constellation, enabling both simplified phase recovery in coherent communication systems and enhanced time-frequency properties in multicarrier frameworks. While Offset-QAM-16 has been leveraged in high-speed optical interconnects using microring modulators for hardware efficiency and power reduction, it also plays a critical role in filter-bank multicarrier (FBMC) systems, such as OFDM/OQAM, for wireless and wireline channel environments. This article presents a comprehensive overview of Offset-QAM-16, spanning its core principles, transmitter realization, performance characteristics, phase recovery advantages, and experimental integration in silicon photonics and multicarrier architectures.

1. Offset-QAM-16: Principles and Constellation Structure

Offset-QAM-16 builds on the conventional 16-QAM constellation, typically arranged as a $4\times4$ Cartesian grid with points at $I, Q \in \{\pm1, \pm3\}$ (normalized units), centered around the origin. In Offset-QAM-16, the entire constellation is displaced from the origin by a real, positive constant $\alpha$ in both in-phase (I) and quadrature (Q) components:

$$I \in \{\pm(2a+1) + \alpha\}, \quad Q \in \{\pm(2b+1) + \alpha\}, \quad a,b \in \{ -1, 0, 1, 2 \}$$

A typical value used is $\alpha \approx 0.42 E_\mathrm{in}$, where $E_\mathrm{in}$ denotes the input electric field amplitude, ensuring that no constellation point sits at the origin (Sturm et al., 13 Jun 2025). This geometric shift yields two principal benefits: (1) improved carrier phase estimation—since small, common-mode phase drifts do not result in symbol overlap at the origin, and (2) support for simple thresholding at the receiver, as the boundaries between adjacent symbol levels remain only shifted, not reconfigured.

In multicarrier applications such as OFDM/OQAM (also known as FBMC/OQAM), Offset-QAM is realized by splitting complex QAM symbols into real-valued components and staggering these in time by $T/2$ to maintain real-field orthogonality between overlapped, filter-shaped symbols (Afrasiabi-Gorgani, 2017).

2. Transmitter Architectures: Microring Resonator-Based and Multicarrier Processing

For high-speed optical interconnects, Offset-QAM-16 transmitters exploit the phase-constant amplitude modulation capability of frequency-detuned RAMZI modulators constructed from silicon microring resonators (MRMs). Each arm of a Mach–Zehnder interferometer (MZI) contains an MRM, thermally tuned to opposite sides of the laser wavelength. Differential voltage swings induce phase shifts $\phi_X$ of equal magnitude but opposite direction, resulting in output fields:

$$E_\mathrm{out} = \frac{1}{\sqrt{2}} [A_T e^{i\phi_T} + A_B e^{i(\phi_B - \phi_{\mathrm{PS}})}]$$

where $$\phi_\mathrm{out} = \arg(E_\mathrm{out}) = -\phi_\mathrm{PS}$$ is held constant by an interferometric phase shifter, decoupling the output amplitude and phase (Sturm et al., 13 Jun 2025). The drive voltages are calibrated to produce four equispaced field levels per I and Q branch with a normalized amplitude contrast $I, Q \in \{\pm1, \pm3\}$0 and phase error $I, Q \in \{\pm1, \pm3\}$1. The offset in symbol positioning translates directly to the electrical drive conditions in each quadrature.

In OFDM/OQAM, the Offset-QAM mapping process comprises:

• Splitting complex $I, Q \in \{\pm1, \pm3\}$2-QAM symbols $I, Q \in \{\pm1, \pm3\}$3 into real components $I, Q \in \{\pm1, \pm3\}$4, $I, Q \in \{\pm1, \pm3\}$5;
• Staggering and phase-rotating these real symbols across time-frequency slots to meet real-orthogonality (Afrasiabi-Gorgani, 2017).

3. Signal Processing: Phase Recovery and Receiver Decision

The origin shift in Offset-QAM-16 directly benefits analog or DSP-free carrier phase recovery (CPR) methodologies. In conventional QAM-16, Costas loop mechanisms struggle with phase estimation due to the presence of points at the origin, which are highly susceptible to phase noise and ambiguous detection. In Offset-QAM-16, the absence of a central constellation point ensures that residual phase rotations result in rotated constellations with non-overlapping decision regions, permitting robust CPR via a simple analog feedback loop that leverages the known offset symmetry (Sturm et al., 13 Jun 2025). The resulting CPR scheme is agnostic to modulation order and data rate, offering scalability.

Receiver hard-decision thresholds are placed at midpoints offset in accordance with $I, Q \in \{\pm1, \pm3\}$6; the comparators, therefore, need no additional complexity compared to standard QAM.

For OFDM/OQAM, receiver processing leverages an analysis filter bank and compensates per-carrier, per-symbol phase rotation, before separating even-indexed symbols (real part) and odd-indexed symbols (imaginary part), enabling regeneration of the original complex symbols (Afrasiabi-Gorgani, 2017).

4. Performance Metrics: Contrast, BER, Spectral Efficiency

In MRM-based Offset-QAM-16, the electric field amplitude contrast per branch is defined as:

$I, Q \in \{\pm1, \pm3\}$7

PDK simulations confirm $I, Q \in \{\pm1, \pm3\}$8 is achievable for all four symbol voltages, maintaining phase fluctuations within $I, Q \in \{\pm1, \pm3\}$9 (Sturm et al., 13 Jun 2025). Bit error rate (BER) versus optical signal-to-noise ratio (OSNR) follows standard QAM-16 analytics, with observed pre-FEC BER of $\alpha$0 at a total laser power of 9.65 dBm for 400 Gb/s operation—comparable to Mach–Zehnder-based QAM-16 links but in 10–100× less silicon area.

OFDM/OQAM with 16-QAM achieves a net spectral efficiency of $\alpha$1 bits/s/Hz without cyclic prefix, surpassing CP-OFDM by $\alpha$2–$\alpha$3 in raw efficiency. The use of well-localized prototype filters (e.g., PHYDYAS with $\alpha$4) ensures a signal-to-interference ratio (SIR) between $\alpha$5–$\alpha$6 dB under ideal conditions. Table 1 summarizes relative metrics:

Platform/Metric	MRM Offset-QAM-16 (Sturm et al., 13 Jun 2025)	OFDM/OQAM (Afrasiabi-Gorgani, 2017)
Electrical contrast	$\alpha$7	Not directly applicable
Pre-FEC BER	$\alpha$8 @ 9.65 dBm (400 Gb/s)	$\alpha$9 at practical SIR
Spectral efficiency	$$I \in \{\pm(2a+1) + \alpha\}, \quad Q \in \{\pm(2b+1) + \alpha\}, \quad a,b \in \{ -1, 0, 1, 2 \}$$0 bits/QAM symbol	$$I \in \{\pm(2a+1) + \alpha\}, \quad Q \in \{\pm(2b+1) + \alpha\}, \quad a,b \in \{ -1, 0, 1, 2 \}$$1 bits/s/Hz (no CP)
SIR	OSNR-limited	$$I \in \{\pm(2a+1) + \alpha\}, \quad Q \in \{\pm(2b+1) + \alpha\}, \quad a,b \in \{ -1, 0, 1, 2 \}$$2–$$I \in \{\pm(2a+1) + \alpha\}, \quad Q \in \{\pm(2b+1) + \alpha\}, \quad a,b \in \{ -1, 0, 1, 2 \}$$3 dB

5. Implementation and Experimental Validation

Experimental demonstration of Offset-QAM-16 modulation has been conducted in the context of silicon photonic transmitters employing the GlobalFoundries 45 nm PDK. A proof-of-concept Offset-QAM-4 implementation at 25 Gb/s validated the calibration protocol and thermal stability of the MRM RAMZI building blocks. The setup included wire-bonded photonic integrated circuits, on-chip CMOS drivers, and external pattern/BERT generators (Sturm et al., 13 Jun 2025). The full Offset-QAM-16 scaling utilizes two parallel RAMZI branches for I and Q with a fixed $$I \in \{\pm(2a+1) + \alpha\}, \quad Q \in \{\pm(2b+1) + \alpha\}, \quad a,b \in \{ -1, 0, 1, 2 \}$$4 phase bias; the same drive calibration naturally extends from QAM-4 to QAM-16 in this architecture.

Thermal analysis demonstrated reliable resonance biasing up to per-ring input optical powers of –4.1 dBm and –2.1 dBm (for 200 and 400 Gb/s, respectively), with no observed bistability or metastable switching under dynamic data operation.

In multicarrier transmission, Offset-QAM is implemented in discrete-time by insertion of the required phase rotation, time-staggering, and polyphase filtering with the selected prototype filter; practical FPGA and ASIC implementations for OFDM/OQAM commonly use these principles (Afrasiabi-Gorgani, 2017).

6. Application Domains and Technological Impact

Offset-QAM-16 modulation is integral to the development of ultra-high-bandwidth, low-area, and low-power optical interconnects for data center environments, particularly in co-packaged optics (CPO) platforms targeting dense GPU and switch integration (Sturm et al., 13 Jun 2025). The simplified phase recovery and hardware efficiency arising from the offset geometry directly address scaling challenges for next-generation silicon photonics and coherent-lite pluggable transceivers.

Parallel advances in FBMC/OQAM employing Offset-QAM-16 yield improved spectral containment and asynchronous access properties for wireless and wireline multiuser scenarios. The lack of a cyclic prefix, enhanced SIR, and robustness to narrowband interference are distinguishing features in multicarrier systems (Afrasiabi-Gorgani, 2017).

A plausible implication is that Offset-QAM-16 will increasingly be adopted in scenarios demanding both stringent power budgets and high spectral efficiency, leveraging its architecture-agnostic support for advanced phase recovery and compatibility with emerging photonic and multicarrier platforms.