A conventional signal, inside a patented spreading process.
MTF stands for mask-matched, time-limited coding with finite access time. The whole family is one patented core configured three ways. Here is what each stage does, how the three configurations differ, and what is already running in silicon.
Five stages. Two of them are ours.
MTF inserts its processing between channel coding and the power amplifier. The diagram reads left to right on transmit and right to left on receive.
Illustrative. The narrowband segment and optional wideband segment both sit under the regulatory mask (dashed).
Standard modulation and FEC.
Identical framing, code rate and overhead to a baseline system, so every comparison is like for like. Your existing coding and MAC stay where they are.
Mask-matched linear spreading.
The patented core applies the MTF spreading code and shapes the spectrum into up to two segments, a narrowband segment and a wideband segment. Both are designed to meet the spectral mask of the jurisdiction, which is the "M" in MTF.
Reach uses the narrowband segment alone. Spectra and Velocity use both, which is where their bandwidth efficiency and throughput come from.
Constant envelope and SNR gain.
The second patented stage gives Reach and Spectra their defining properties: a constant envelope (0 dB PAPR) and an SNR gain of up to 7.3 dB without heavy forward error correction.
The two stages are inseparable. The linear stage makes the signal mask-compliant; the non-linear stage extracts the gains. Velocity leaves this stage out and trades those gains for raw rate.
The amplifier runs flat out.
A constant envelope means no back-off. The power amplifier runs fully saturated at maximum efficiency, which improves EIRP for the same bit rate. Velocity is linear and uses conventional back-off, as OFDM does.
Iterative interference cancellation.
The spreading deliberately overlaps symbols. An MMSE receiver with successive interference cancellation separates them, estimating and removing residual inter-symbol interference over several passes until it converges on a clean estimate.
Velocity, with no non-linear stage to invert, uses a linear equaliser matched to the transmit spreading.
Choose the axis you need to win.
One configuration per application. All three share MTF Core and a drop-in PHY.
Power efficiency and reach.
Reach operates at a very low Eb/N0, so it needs the least transmit power for a given link, or goes the furthest for a given power. Its spectrum is a single narrowband segment that fills only the allocated channel, so the transmitter looks like any other radio to the regulator. Out-of-band emissions are filtered to the mask.
| Eb/N0 at BER 10⁻³ | ≈3 dB, uncoded Sim |
|---|---|
| vs theoretical BPSK | about 4 dB better at the same BER |
| Envelope | Constant, 0 dB PAPR |
| Amplifier efficiency | up to ~4× that of conventional OFDM |
| Spectrum | Narrowband segment only |
| Multiple access | Yes |
| Hardware | Measured Transmitter and full receiver running 100% in FPGA; image available |
Bandwidth efficiency.
Spectra packs a high data rate into a narrow occupied band, so it delivers more throughput per allocated hertz. Like Reach it has a constant envelope, so the amplifier still runs saturated. Error correction is very light: Reed–Solomon RS(255, 247), 3.2% overhead.
| Rate in an 8 MHz channel | ≈1.7 Gbps Sim |
|---|---|
| Eb/N0 required | ≈9 dB |
| Envelope | Constant, 0 dB PAPR |
| Spectrum | Narrowband + wideband segments |
| Impairments modelled | CFO, sampling-clock offset, I/Q imbalance and more |
| Multiple access | No, single link |
| Hardware | Measured Proven on RFSoC hardware and in emulation. In progress FPGA image |
Raw throughput.
Velocity keeps Spectra's narrowband spreading and adds higher-order modulation. It gives up the constant envelope and the SNR gain in exchange for the highest raw bit rate per channel of the three, for short-range backhaul, point-to-point links and high-SNR fixed installations.
| Rate in an 8 MHz channel | ≈22 Gbps at 64-QAM Sim |
|---|---|
| Eb/N0 required | ≈22 dB at 64-QAM |
| Modulation | Configurable; BPSK to 64-QAM in FPGA today |
| Envelope | Linear, with conventional back-off |
| Receiver | Linear equaliser matched to the spreading |
| Multiple access | Yes |
| Hardware | Measured Transmitter and receiver running 100% in FPGA; 16- and 64-QAM decoded in FPGA; image available |
Same 8 MHz channel. Four waveforms.
Each configuration against a baseline OFDM system with the same coding and overhead. Simulated
| Waveform | Bit rate in 8 MHz | Eb/N0 needed | Constant envelope | Multiple access |
|---|---|---|---|---|
| MTF Reach | 2 Mbps | 3 dB | Yes | Yes |
| MTF Spectra | 1.7 Gbps | 9 dB | Yes | No |
| MTF Velocity | 22 Gbps | 22 dB | No | Yes |
| Conventional OFDM | 80 Mbps | 28 dB | No | Yes |
Bit-rate bars use a log scale. Velocity is quoted at 64-QAM.
Hover or tap the chart for values. Points are read from MTFComm simulation results; values between points are interpolated on a log scale. Reach is shown uncoded; Spectra uses RS(255, 247).
BER data points as a table
| Curve | Eb/N0 (dB) → BER |
|---|---|
| MTF Reach | 1.9 dB → 5e-2 · 2.1 dB → 2e-2 · 2.3 dB → 1e-2 · 2.6 dB → 2.2e-3 · 2.9 dB → 1.1e-3 |
| MTF Spectra | 7.6 dB → 1.1e-1 · 8.2 dB → 1.4e-2 · 8.7 dB → 7e-3 · 9.2 dB → 2.2e-3 · 9.7 dB → 3e-4 |
| Velocity · BPSK | 10.6 dB → 1.6e-2 · 11.6 dB → 1e-2 · 12.7 dB → 4e-3 · 13.7 dB → 3.3e-4 |
| Velocity · 16-QAM | 14.3 dB → 1.5e-2 · 15.3 dB → 1.2e-2 · 16.2 dB → 3.6e-3 · 17.2 dB → 1.2e-3 |
| Velocity · 64-QAM | 18.2 dB → 1.9e-2 · 19.2 dB → 1.2e-2 · 20.2 dB → 5e-3 · 21.2 dB → 2.6e-3 · 22.2 dB → 2e-5 |
| Theoretical BPSK | 0.5 · erfc(√(Eb/N0)); 1e-3 at 6.8 dB, 1e-5 at 9.6 dB |
Simulation, AWGN. Points read from the MTFComm Technology Briefing (August 2026).
Not a simulation. Running in silicon.
Reach and Velocity run 100% in FPGA fabric, transmitter and receiver, with no host processor in the loop. Their FPGA images are available to partners for evaluation.
| Platform | AMD Zynq UltraScale+ RFSoC ZCU208 evaluation kit |
|---|---|
| Device | XCZU48DR, Gen 3 RF data converters |
| Converter rate | 4.42 GS/s (validated to 4.8 GS/s) |
| Fabric clock | 276.48 MHz |
| Test link | Two boards, one transmitting and one receiving, over a board-to-board RF link |
MTF Reach 100% FPGA
Transmit and receive images complete. The full receiver runs in the fabric with no host processor, keeping up at full duty with no dropped frames.
MTF Velocity 100% FPGA
Transmit and receive images complete from BPSK to 64-QAM. 16- and 64-QAM decoded in the FPGA.
MTF Spectra FPGA in progress
Proven on RFSoC hardware with host-side processing, and in full emulation. The FPGA image is being written, targeting AMD's next-generation Versal RF devices.
Built to become a chip
One RTL codebase generates all three configurations and is written to port to a single custom ASIC.
Evaluate MTF on your bench.
FPGA images for Reach and Velocity, the technical briefing and full patent documentation are available to qualified partners under NDA.