The 100G coherent revolution
The most critical factor driving the architecture of long-haul DWDM networks is the cost per bit transmitted per kilometer traversed – or cost/bit/km. While each successive generation of DWDM technology has increased bandwidth, transmitting successively higher rates over long distances while decreasing cost has often proved challenging.
Introduced around 2010, 100G coherent technology represented a sea change in optical technology and its concomitant network architectures. Coherent 100G provided 10X the bandwidth of 10G and 2.5X that of 40G. Its use of digital signal processing resulted in exceptional long-haul performance immune to the linear and non-linear effects that were such a severe challenge for the multiple generations of DWDM that relied on direct detection. It’s no surprise that 100G quickly resulted in a far lower cost/bit/km than 10G and 40G and, within five years, became the dominant long-haul technology. Today, new long-haul networks are built almost exclusively around 100G coherent technologies.
While 100G has become the dominant long-haul line rate, the vast majority of clients interface to long-haul networks at 10G. The rapid growth of mobile broadband, IP video, and cloud services is increasing network bandwidth to the point where 100G is becoming a common service currency between services infrastructure and transport networks. Network operators need a generational leap in scale to cost effectively deliver 100G connectivity and avoid exhausting the capacity of their transmission systems.
Is it enough? Looking beyond 100G
Like the crucial metric of cost/bit/km, spectral efficiency (translating to system capacity) is a factor in urgent need of consideration. It’s still possible to transport 100G clients over 100G DWDM transport wavelengths. But this approach has serious implications relative to network capacity, particularly in long-haul, ultra-long-haul, and subsea networks, where fiber is scarce and expensive.
Most existing long-haul networks have an optical fiber capacity of roughly 10 terabits per second, a simple multiplication of approximately 100 channels at 100 gigabits per second. This capacity is halved for ultra-long-haul and subsea networks, where 50G channels are the norm (and where extreme distance requirements demand a more robust, and hence lower-rate, modulation format). Now that 100G services crisscross the globe, these capacities must grow beyond their current levels, driven by DWDM wavelengths that deliver higher rates, lower costs, and exceptional optical performance.
200G to the rescue – but what about long haul?
This transformation has already begun. In 2014, Nokia became the first to ship a 200G single-carrier DWDM interface, enabling network operators to double the transport capacity of their existing DWDM infrastructures.
The first-generation Nokia Photonic Service Engine coherent digital signal processor (DSP) allowed a single line card to operate in the well-established mode of DP-QPSK for 100G wavelengths, as well as in DP-16QAM for 200G. The use of 16QAM increased the number of bits encoded within one analog symbol from two to four. It doubled capacity and spectral efficiency while operating within the capabilities of then-state-of-the-art DSP silicon, around 33 gigabaud.
As with so many interactions with the physical world, these gains have come at a price. With more bits, and thus more possible symbols to be discerned within the same Euclidean space of phase and amplitude, a higher optical signal-to-noise ratio (OSNR) is required to achieve the near error-free transmission expected of optical transport systems. A higher OSNR requirement dictates fewer noise-contributing amplifiers, fewer amplified spans and, ultimately, shorter achievable distances. Whereas 100G coherent DWDM interfaces are capable of unregenerated reach in excess of 3000 km, 200G 16QAM interfaces top out at approximately 1000 km.
Expanding long-haul capacity
The increase in system capacity brought about by 16QAM is welcome because the distribution of content and cloud services into geographically dispersed data centers is causing metro traffic to grow at nearly twice the rate of long-haul traffic. However, the unique economics of long-haul networking demand a similar step-function increase in system capacity to address inter-city, transcontinental, and subsea routes. These routes are also seeing significant bandwidth growth. They would benefit from higher capacity and reduced cost/bit/km. Realizing these benefits demands a new generation of coherent technology.
The Nokia Photonic Service Engine 2 Super Coherent (PSE-2s) DSP is designed to address this problem. Using a state-of-the-art silicon process, the PSE-2s packs 1.4 billion transistors into the world’s most sophisticated and highly integrated coherent DSP. The PSE-2s can select from multiple modulation formats and operate at 33 or 44 gigabaud. It supports several unique operating modes while continuing to interwork with existing deployments.
200G for the long haul
By combining high order modulation with high baud rate transmission, the PSE-2s enables an 8QAM modulation format that operates at 200G data rates. Encoding three bits per symbol, 8QAM improves on the spectral efficiency of 100G QPSK while avoiding the constrained performance of 200G transmission based on 16QAM. It brings true long-haul performance to 200G wavelengths over distances of up to 2000 km (Figure 1). With this kind of reach, operators can build long-haul networks economically, using 200G wavelengths almost exclusively.
Figure 1: Nokia 8QAM brings long-haul performance to 200G wavelengths
Compared to long-haul networks built with 100G or alternative 8QAM solutions running at 150G, the Nokia 200G 8QAM uses fewer line cards, offers more spectral efficiency, and delivers the lowest cost/bit/kilometer. It also offers better alignment with the 100G transport services that customers are starting to demand.
Ultra-long-haul 100G
100G coherent interfaces are capable of achieving distances of 3000 km or more. But some transcontinental routes and most subsea routes exceed this distance. They require expensive regeneration or less efficient and more costly binary phase shift keying (BPSK) modulation.
By reducing the number of bits encoded per symbol from two to one, BPSK signals can support unregenerated reaches two to three times greater than those that can be achieved with quadrature phase shift keying (QPSK). However, BPSK requires two carriers to transmit 100G payloads. This increases system complexity and reduces spectral efficiency.
The PSE-2s combines high baud rate capabilities with an advanced version of QPSK called set partition QPSK (SP-QPSK) to enable the first single-carrier QPSK-modulated wavelengths capable of ultra-long haul distances (Figure 2).
Figure 2: High baud rates and SP-QPSK stretch 100G reach beyond 5000 km
Set partition modulation encodes approximately 1.5 bits per symbol while operating at a baud rate 33 percent higher than standard QPSK. It supports the implementation of a sophisticated algorithm that reduces inter-symbol interference and increases performance by coordinating adjacent symbols across the time domain. SP-QPSK improves OSNR by 2.5dB, which translates to achievable distances greater than 5000 km – nearly double what is possible with standard QPSK. This performance can eliminate the need for terrestrial regeneration and support direct operation over many subsea transmission systems.
Summary
100G transmission revolutionized long-haul DWDM transport by dramatically increasing capacity and lowering cost/bit/km. However, these advances are no longer sufficient to keep up with surging bandwidth demand. Platforms powered by the Nokia PSE-2s double long-haul and ultra-long-haul capacities, enabling operators to continue decreasing the cost per bit transmitted while expanding the capacity of their long-haul DWDM systems.