The Need for Flexibility: From Rigid Highways to Smart Roads
Traditional operate on a fixed grid. Imagine this as a highway with a set number of lanes, all of exactly the same width (e.g., 50 GHz). A small motorcycle (a low-data-rate service) is forced to occupy a full lane, wasting most of the space. A very wide truck (a high-data-rate service) might not fit in a single lane and would need to be split across two, which is inefficient.
This "one-size-fits-all" approach leads to significant waste of the most precious resource in an optical network: the spectrum. To solve this, the concept of the Flexible Grid was introduced, transforming the rigid light highway into a smart, dynamic road system where lane widths can be precisely tailored to the traffic they carry. This is the core technology that enables Elastic Optical Networks (EON).
Flexible-Grid Parameters
In a flexible grid, an optical channel is described by its nominal central frequency and frequency-slot width. This allows the assigned spectrum range to be matched to the requirements of a particular signal.
Frequency Slot
According to ITU-T G.694.1, a frequency slot is defined by two independent parameters: its nominal central frequency and slot width .
- Slot width: , where is a positive integer. The slot-width granularity is therefore 12.5 GHz.
- Nominal central frequency: , where is an integer. The 6.25 GHz spacing applies to possible central-frequency positions, not to the slot width.
Constructing an Optical Channel
With the flexible grid, an optical channel is no longer a fixed entity but a dynamic construct created on demand. It is defined by allocating a specific number of adjacent frequency slots to a connection.
- Contiguous Allocation: The frequency slots assigned to a single connection must be , meaning they form an unbroken, continuous block of spectrum. You can't assign slots 1, 2, and 5 to the same channel while skipping 3 and 4.
- Variable Width: The total width (bandwidth) of the channel is determined by the number of slots () it occupies.
- Guard Bands: To prevent interference between adjacent channels, a small gap, known as a , is required. This is typically achieved by leaving one or more frequency slots empty between the allocated blocks of two different connections.
The Rules of the Road: Spectrum Constraints
The flexibility of the grid comes with two critical rules, or constraints, that must be respected when establishing a connection across the network.
1. The Spectrum Contiguity Constraint
As mentioned above, the frequency slots allocated to a single connection on any given fiber link must form a single, solid, unbroken block. This is a physical requirement of the underlying optical components (like switches and filters) that are designed to handle continuous slices of spectrum. Think of it as a train: all the carriages must be coupled together; you cannot have a gap in the middle.
2. The Spectrum Continuity Constraint
This is a more challenging and impactful constraint. When an optical connection travels across multiple fiber links (e.g., from Node A to B, then B to C, then C to D), it must occupy the exact same block of frequency slots on every single link along its entire path.
Imagine driving from London to Rome. The Spectrum Continuity Constraint is like a rule stating that if you start in lane 3 on the M25 motorway, you must remain in lane 3 on every subsequent motorway segment all the way to your destination. You cannot switch lanes.
Why this rule? This constraint exists because most optical network nodes currently lack the ability to perform efficient and cost-effective all-optical wavelength conversion. An optical switch can redirect a block of light from an input to an output, but it cannot easily change its color (its position in the frequency spectrum). Finding a common, contiguous, and continuous block of free slots across a long, multi-hop path is the central challenge of the problem and can lead to a phenomenon called spectrum fragmentation.