At the whole-link level you need a model that lets you analyse the behaviour of the system as a whole. Instead of solving the electromagnetic field in a single element in detail, you build a model from connected blocks representing the transmitter, fiber, amplifiers, filters, receiver, and signal-processing circuits, and then check the effect the whole path has on transmission quality.
One tool intended for this task is OptiSystem from Optiwave — an environment for designing and simulating optical systems, covering both individual components and subsystems and complete transmission links. This article closes the series: PIC — the photonic integrated circuit describes the integrated-circuit level; here we move to the whole-system level.
From component to system
In photonics simulation it is worth distinguishing three levels of description.
- Component — how does a specific element behave (laser, modulator, coupler, filter, waveguide, fiber span)? We are interested in the physical phenomena inside that element.
- PIC — how do the elements of an integrated photonic circuit work together (interferometers, couplers, modulators, waveguides)?
- System — does the whole transmission path meet the signal-quality requirements? We no longer want to analyse the field in every fiber segment — we want to know whether the information will get through the whole link with adequate quality.
- Component — how does a specific element behave (laser, modulator, coupler, filter, waveguide, fiber span)? We are interested in the physical phenomena inside that element.
- PIC — how do the elements of an integrated photonic circuit work together (interferometers, couplers, modulators, waveguides)?
- System — does the whole transmission path meet the signal-quality requirements? We no longer want to analyse the field in every fiber segment — we want to know whether the information will get through the whole link with adequate quality.
This hierarchy is directly present in OptiSystem: you can work at the system, subsystem, and component level, and a subsystem can be built from a group of components or other subsystems. The whole process can be framed as: component → PIC / subsystem → system → transmission-quality parameter. And that parameter can be, among others, BER.
What are we actually simulating?
The simplest fiber-optic link can be represented as a chain of blocks: data → transmitter → fiber → receiver → decision → BER. In a real system there will of course be more blocks.
The transmitter can contain a bit-sequence generator, an optical source, a modulator, and driving and electrical signal-processing circuits. The optical path can include the fiber, attenuation, dispersion, nonlinearities, filters, optical amplifiers, and WDM multiplexers and demultiplexers. The receiver can contain a photodetector, an amplifier, an electrical filter, clock recovery, a detection and decision circuit, and a BER analyzer.
OptiSystem allows modelling such systems in both the time and frequency domains; the documentation lists, among others, WDM systems, single-mode and multi-mode transmission, EDFA/SOA/Raman amplifiers, direct and coherent transmitters and receivers, and various modulation formats. This is an important difference from simulating a single component: we no longer ask „what happens in this element?”, but „what happens to the information once it passes through the whole path?”.
The block-based approach — that is, system-level simulation
The most convenient way to build such a model is the block-based approach. Each block represents a specific system function; the blocks are connected, and the signal passes through the successive stages of the path — e.g.: PRBS → laser → modulator → fiber → EDFA → filter → photodetector → electrical filter → decision → BER Analyzer.
This does not mean every block is a simple „box” with one parameter — appropriate physical properties and mathematical models can be included inside. From the system designer's point of view, the important thing is that these models can be connected into one transmission path, and then you can change, for example, the fiber length, transmitter power, modulator parameters, modulation format, attenuation, dispersion, amplifier parameters, noise level, or filter characteristics — and check how the change affects the result of the whole link. OptiSystem is a multi-level tool: alongside a component library, it provides the creation of subsystems and organising a project in a hierarchical structure.
The transmitter: from bits to light
The simulation starts with information — in the simplest case a bit sequence from a pseudo-random generator. The information is represented as an electrical signal driving the optical transmitter; then come the light source (e.g. a laser) and a modulator. Depending on the architecture, we may have direct or external modulation, and in more advanced systems the model may include multi-level or coherent modulation. Among the supported formats, Optiwave lists NRZ, RZ, DPSK, QPSK, DP-QPSK, and QAM.
At the transmitter output we are no longer interested only in the optical power, but in the encoded information. This is key, because later distortions have to be assessed in the context of the ability to recover that information.
The fiber: the signal starts to change
After the transmitter, the signal enters the transmission path. The simplest model is a single fiber span; in a real link the signal can experience, among others, attenuation, chromatic dispersion, nonlinear effects, polarisation-related changes, and the influence of amplifiers and other path elements.
Optiwave shows examples of system design that take into account, among others, fiber length, attenuation, dispersion, and nonlinearities. This is where the advantage of system-level simulation is clearly visible: we can not only compute the parameters of the fiber itself, but also check how its properties affect the information-carrying signal. A concrete example — Verifying a DWDM link.
An amplifier does not „fix” everything
In a long link there are optical amplifiers. Their job is to compensate for power loss, but increasing the signal power does not automatically restore its original quality — an amplifier can add its own noise. That is why, in a system-level simulation, you have to look not only at how much power the signal has, but also at the signal-to-noise ratio.
That is exactly why one of the important analysis results is OSNR (Optical Signal-to-Noise Ratio). OptiSystem has tools for analysing signal power and OSNR, including per WDM channel.
The receiver: this is where the real test begins
At the end of the optical path is the receiver. In the simplest case a photodetector converts the optical signal into an electrical one; the signal can then be filtered, amplified, and subjected to detection and decision. Only here can you ask the fundamental question: is the receiver able to correctly recover the transmitted bits?
The optical power at the receiver input alone does not answer this — two signals can have similar power and yet differ greatly in transmission quality. That is why, in a system-level simulation, we need appropriate visualisers and analyzers.
The eye diagram — the first picture of signal quality
One of the most characteristic results of a link simulation is the eye diagram. It is formed by overlaying successive waveform segments corresponding to bit periods and gives an intuitive picture of how much the signal has been distorted:
- a larger eye opening means a larger decision margin;
- a closing eye indicates increasing distortion or noise;
- horizontal eye closure is associated with timing problems;
- vertical closure limits the amplitude margin.
- a larger eye opening means a larger decision margin;
- a closing eye indicates increasing distortion or noise;
- horizontal eye closure is associated with timing problems;
- vertical closure limits the amplitude margin.
OptiSystem provides the eye diagram as part of the BER analysis — the BER Analyzer shows, among others, the eye diagram, Q-factor, minimum BER, decision threshold, and decision instant. The eye diagram is therefore more than an impressive chart: it is a visual diagnostic of the transmission path's quality.
Q-factor — a numerical description of quality
The eye diagram is useful, but a designer often also needs a number — one of them is the Q-factor. In the classical case of binary transmission, with a statistical description of the signal levels, Q can be related to the distance between the distributions of the „1” and „0” levels, referred to their standard deviations; for appropriate assumptions, Q can then be related to an estimated BER.
In practice, however, you have to be careful about automatically interpreting Q as a universal substitute for BER — the result depends on the model used and the assumptions about signal, noise, and detection. Optiwave states that OptiSystem uses numerical analysis or semi-analytical techniques to determine BER and Q-factor in systems limited by intersymbol interference (ISI) and noise. The Q-factor is a measure of quality, BER is a measure of transmission errors — they are not two names for the same result.
BER — the system's most important answer
BER (Bit Error Rate) answers the question: how often does the receiver make a wrong bit decision? In the most direct terms, BER = number of incorrectly received bits / number of all received bits. In simulation there are different ways to obtain this result — which is especially important in OptiSystem.
BER Analyzer
The BER Analyzer can be used to estimate BER. In an Optiwave example for NRZ transmission, the estimation is done based on a Gaussian model and a short bit sequence — an analytical/statistical approach, not counting every erroneous bit in a very long transmission.
BER Test Set
The second option is the BER Test Set, in which errors are actually counted by comparing the corresponding sequences. This requires a sufficiently large number of transmitted bits for the result to be statistically meaningful. If a designer wants to quickly assess system quality, estimation is very useful; if they want to count errors experimentally, they need a long sequence and a properly configured detection path.
Why does the simulation length matter?
BER is a statistical quantity. If a simulation transmits only a small number of bits and no error is observed, this does not mean the real BER is zero — it may simply mean no error occurred in the sample. The smaller the BER we want to measure reliably by directly counting errors, the more bits we need.
That is why a system-level simulation has to account for a trade-off: statistical accuracy ↔ computation time. OptiSystem allows working with different signal representations and time-driven simulation; with sample-by-sample simulation, each sample may require a separate calculation, which increases the simulation time.
OSNR — when we care not only about time, but about the spectrum
The eye diagram and BER describe transmission quality mainly from the side of the time-domain signal and the bit decision. In optical systems, however, we often also need spectral information — this is where OSNR (Optical Signal-to-Noise Ratio) comes in, describing the relation between the optical signal power and the optical noise power in a defined reference bandwidth.
OSNR is especially important in systems with optical amplifiers and WDM transmission, where you have to analyse not only a single time waveform but also the spectrum of many channels. OptiSystem provides spectral analysis and tools that yield, among others, signal power, gain, noise figure, and OSNR per WDM channel. The same system can thus be viewed from several sides: time → eye diagram, quality statistics → Q-factor, errors → BER, spectrum → OSNR. None of these results replaces the others — together they give a fuller picture of the link's behaviour.
What do we actually get from a system-level simulation?
A well-built optical link simulation answers far more questions than just „does the signal reach the receiver?”. A typical set of results:
| Result | What does it tell you? |
|---|---|
| Eye diagram | what the signal's timing and amplitude margins look like |
| Q-factor | a numerical description of the quality of level/state separation under the assumptions made |
| BER | how often wrong bit decisions occur, or their estimate |
| OSNR | the ratio of optical signal to noise in a defined bandwidth |
| Optical spectrum | how the signal spectrum changes |
| Power | how the power changes at successive points along the path |
| Constellation diagram | symbol quality in systems with multi-level / coherent modulation |
| Penalties | the effect of specific degradations on system performance |
Among the system-analysis results, Optiwave lists, among others, the eye diagram, Q-factor, BER, signal power, OSNR, polarisation states, constellation diagrams, and linear and nonlinear penalties.
The most interesting part: parametrising the system
Being able to run one model is only the beginning. The real value of system-level simulation appears when we can change parameters and observe the result — e.g. fiber length → BER, transmitter power → BER, OSNR → BER, fiber length → Q-factor, distance → eye diagram. In this way a single simulation yields a dependency analysis.
OptiSystem includes parametric-analysis mechanisms; the vendor's materials include, among others, a tutorial on BER × input power analysis. In practice, a designer is not looking for just one „good” operating point — they are looking for a design margin. More on choosing the level and method of simulation → Methods for modelling optical phenomena.
When is a component simulation not enough?
Suppose we know the fiber parameters and have analysed its attenuation, dispersion, and propagation properties in detail. We still do not know whether:
- the transmitter has enough power;
- the modulator operates at the right point;
- the amplifier does not degrade OSNR more than assumed;
- the receiver has enough margin;
- the filter does not cause excessive eye closure;
- the whole combination of elements provides the required BER.
These are system questions. A component accounts for the local behaviour of an element, a system — for the cumulative effect of the whole path.
- the transmitter has enough power;
- the modulator operates at the right point;
- the amplifier does not degrade OSNR more than assumed;
- the receiver has enough margin;
- the filter does not cause excessive eye closure;
- the whole combination of elements provides the required BER.
These are system questions. A component accounts for the local behaviour of an element, a system — for the cumulative effect of the whole path.
Not everything should be simulated at one level
If the problem concerns the waveguide geometry, we do not start with a full link simulation. If it concerns the characteristics of a single modulator, we analyse the modulator first. If we want to check whether the whole system reaches the required BER, we build a system model. The level of detail of the model should match the question we want to answer.
Optiwave separates its environments into tools for different tasks: OptiSystem is for designing and simulating optical systems, while other products in the family — OptiFDTD, OptiBPM, OptiMode, OptiOmega, OptiSPICE — are intended for more detailed analysis of devices and circuits. This is not competition between a „better” and a „worse” simulation, but different questions and different modelling levels.
Where does a PIC fit into this hierarchy?
If a single component is an element and a link is a system, a PIC sits between these levels as an integrated structure of many photonic elements (waveguides, couplers, splitters, interferometers, modulators, detectors, filtering elements). At the PIC level we ask about the functioning of the whole integrated structure; at the system level such a PIC can be treated as a single functional block.
This is an important concept of hierarchical modelling: we do not have to expand the whole internal structure of the circuit every time — if we know its input–output characteristics, we can use it as an element of a larger system. This is exactly what subsystems in OptiSystem are for.
Component → PIC → system
| Level | Question | Typical result |
|---|---|---|
| Component | how does a single element work? (laser, modulator, coupler, fiber, filter) | element characteristics, transmission, field, spectrum, physical parameters |
| PIC | how do the elements of an integrated circuit work together? (interferometer, integrated modulator, signal path) | characteristics of the whole circuit, relationships between ports, spectral and functional response |
| System | will the information get through the whole transmission path? (link, WDM, PON, coherent transmission, EDFA system) | eye diagram, BER, Q-factor, OSNR, power, spectrum, constellation diagram, system margin |
System-level photonics simulation sits at the system level.
OptiSystem as an example of a system-level tool
OptiSystem is an environment designed for exactly this level of analysis — it lets you connect hundreds of components to simulate different types of links and covers the analysis of both individual elements and complete systems. What matters is not only the component models themselves, but the ability to connect them.
This changes how you work. Instead of „let's analyse the fiber” we ask „what will happen to BER if we increase the fiber length from 50 to 100 km?”. Instead of „what is the amplifier's noise level?” — „how does the amplifier's noise affect the OSNR and BER of the whole link?”. Instead of „what does the signal look like after the filter?” — „does the filter cause such eye closure that the receiver no longer meets the requirements?”. That is the difference between component analysis and system analysis.
Simulation does not replace measurement — but it lets you design the experiment better
A system-level simulation is not a magic equivalent of a real link. The result always depends on the models used, the component parameters, the signal representation, the noise assumptions, the sequence length, the receiver settings, and the BER calculation method — so the model should be validated.
Before a physical prototype exists, however, simulation lets you check a huge number of variants: find configurations worth building, reject solutions that obviously fail the requirements, determine the system's sensitivity to individual parameters, and see much earlier where the limitation of the whole link lies.
The most important change in thinking
The greatest value of a system-level simulation is not the BER chart itself, but the ability to move from the question „does this component work?” to the question „does the whole system work as it should?”. A component may work correctly, a PIC may work correctly, and yet the whole link may fail to meet the requirements.
The reason can be the sum of small degradations: attenuation + dispersion + noise + bandwidth limitations + nonlinearities + transmitter imperfection + receiver imperfection. Only a system model lets you see their combined effect.
Summary
Simulating an optical link is the analysis of the whole transmission path — from generating the information, through the transmitter and the optical channel, to the receiver and the bit decision. Its basis is the block-based approach: the successive elements of the system are represented as components or subsystems connected into one model.
In a tool such as OptiSystem you can analyse the system in the time and frequency domains and obtain results such as eye diagram → Q-factor → BER and power → spectrum → OSNR. This lets you not only see that the signal degrades, but determine whether the degradation is still acceptable from the point of view of the whole system. The principle stays simple: a component answers „how does an element work?”, a PIC — „how does an integrated circuit work?”, and a system — „does the information reach the receiver correctly?”. We set out the criteria for choosing a package in the guide How to choose optical simulation software?.