The path to 1Tb/s and beyond datacenter interconnect networks: technologies, components, and subsystems

Spyropoulou, M.; Kanakis, Giannis; Jiao, Yuqing; Calabretta, Nicola; Williams, Kevin; Nodjiadjim, Virginie; Hersent, R.; Konczykowska, A.; Schatz, Richard; Ozoliņš, Oskars; Joharifar, Mahdieh; Zvěřina, Jakub; Zoldak, Martin; Bakopoulos, P.; Patronas, Giannis; Avramopoulos, H.

doi:10.1117/12.2582920

Cited by 10 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Hence, InP DHBTs could bridge the performance gap and empower next generation beyond 1-Tb/s/channel optical transceivers [56], [57], as well as enabling sub-THz power generation [58] for beyond 5G/6G applications. Although, to date, they remain niche-market technologies, regarding communication applications, intense research activities are conducted to combine InP DHBTs' potential with the highmaturity silicon technologies [59], [60], [61], [62].…”

Section: Discussionmentioning

confidence: 99%

InP DHBT Linear Modulator Driver With a 3-Vppd PAM-4 Output Swing at 90 GBaud: From Enhanced Transistor Modeling to Integrated Circuit Design

Hersent,

Johansen,

Nodjiadjim

et al. 2024

IEEE Trans. Microwave Theory Techn.

View full text Add to dashboard Cite

In this article, we report on the modeling, design, and characterization of indium phosphide (InP) double heterojunction bipolar transistor (DHBT) devices and integrated circuits (ICs) for next-generation optical communications. Critical aspects of transistors' modeling and their influence on the IC design are detailed, as well as the design and characterization of a lumped linear modulator driver featuring a 3-Vppd four-level pulse-amplitude modulation (PAM-4) output swing at 90 GBaud (GBd). In particular, we propose an electromagnetic (EM) simulation-based parasitic extraction method of the DHBT access structures, to refine the DHBT and IC performance prediction accuracy. It is shown to provide a better estimation of a canonical cascode gain and µ stability factor at millimeterwave frequencies, as well as a better estimation of the driver IC gain in the 50-110 GHz frequency range. Furthermore, a highfrequency gain boosting (self-peaking) topology, based upon an emitter-degenerated paralleled-transistor cascode configuration, is analyzed using a simplified transistor model and leveraged to enhance the linear driver output-stage gain-bandwidth product with controlled amount of peaking gain. This self-peaking technique is shown to be inherent to cascode structures and can therefore be used with other technologies, with no added design complexity. The driver IC was implemented in a 0.5-µm InP-DHBT technology and features a bandwidth well in excess of 110 GHz, with 13 dB of peaking gain at 95 GHz. Besides,

show abstract

Section: Discussionmentioning

confidence: 99%

InP DHBT Linear Modulator Driver With a 3-Vppd PAM-4 Output Swing at 90 GBaud: From Enhanced Transistor Modeling to Integrated Circuit Design

Hersent,

Johansen,

Nodjiadjim

et al. 2024

IEEE Trans. Microwave Theory Techn.

View full text Add to dashboard Cite

show abstract

“…This is because the CPU-CPU traffic is usually latency-sensitive short packets, and the memory-memory traffic is usually for long running scenarios like direct memory access (DMA) and live virtual machine migration requiring high bandwidth and throughput. The bandwidth of network interconnections in DACON can remarkably scale to higher than 100 Gb/s without modification of the NOS by employing QSFP28DD transceivers as well as wavelength division multiplexing (WDM) and channel bundling technologies [28]. This eliminates and surpasses the performance bottleneck of the interconnections in the server-centric architecture in terms of low bandwidth and high latency.…”

Section: Dacon Architecture and Operationmentioning

confidence: 99%

DACON: a reconfigurable application-centric optical network for disaggregated data center infrastructures [Invited]

Guo

Xue

Yan

et al. 2021

J. Opt. Commun. Netw.

Self Cite

View full text Add to dashboard Cite

To solve the issues of low resource utilization and performance bottleneck in current server-centric data center networks (DCNs), we propose and experimentally demonstrate a disaggregated application-centric optical network (DACON) for data center infrastructures based on hybrid optical switches. DACON achieves flexible provision and dynamic reconfiguration of hardware nodes exploiting the softwared-define networking (SDN) orchestration plane. A four-node SDN-enabled disaggregated prototype is implemented with a field-programmable-gate-array-based controller of hardware nodes and nanosecond optical switches, performing a minimal node-to-node network latency of 378.6 ns and zero packet loss. Based on the unmodified Linux kernel and two different applications (distributed computing and a Memcached database), the application runtime of the disaggregated prototype is investigated and compared with the server-centric architecture. Experimental results show that the disaggregated prototype performs better with Memcached database applications, achieving a 1.46 × faster runtime than the server-centric network at a memory node access ratio of 0.9. Based on the customized control plane orchestrator and dynamic resource reallocation, the node-to-node latency is reduced by 21% when CPU nodes access memory nodes. The scalability of DACON is then numerically assessed based on experimentally measured parameters. Results show that the intra-rack node-to-node latency is less than 404.8 ns with a 6144-node network and memory node access ratio of 0.9. Finally, the cost and power consumption are also studied and compared with current DCN architectures. Results indicate DACON saves 13.4% of the cost of an interconnect network compared with current disaggregated architecture and consumes up to 31.1% less power with respect to server-centric DCN architectures.

show abstract

“…With the ever-increasing demand for data-intensive applications such as high-resolution video streaming and cloud computing, there has been a growing need for high-capacity datacenter interconnection. [1][2][3] Such demands call for bandwidth-efficient, low-latency, and cost-effective short-reach optical transmission systems. Although coherent detection has been extensively employed in long-haul transmission and metro networks due to the high electrical spectral efficiency (ESE) and high receiver sensitivity, it is not favored in modern short-reach works because of its excessive cost and complexity.…”

Section: Introductionmentioning

confidence: 99%

Spectrally Efficient Integrated Silicon Photonic Phase‐Diverse DD Receiver with Near‐Ideal Phase Response for C‐Band DWDM Transmission

Zhang,

Li,

et al. 2023

Laser & Photonics Reviews

View full text Add to dashboard Cite

In this paper, a spectrally efficient silicon photonics phase‐diverse direct‐detection receiver based on asymmetric self‐coherent detection is proposed. The receiver is hardware efficient and composed of only two photodiodes and two analog‐to‐digital channels to accurately recover complex double‐sideband signals by leveraging an optimized integrated silicon racetrack resonator with near‐ideal phase responses. Moreover, the periodicity of the resonator response enables the receiver to support multi‐channel transmission such as dense wavelength‐division multiplexing. The receiver circuit is fabricated and characterized. In 40‐km transmission experiments, a record net electrical spectral efficiency of 7.10 bit s−1 Hz−1 per wavelength and per polarization is achieved, where a net 208‐Gb s−1 32QAM transmission is demonstrated using 29.3‐GHz electrical bandwidth. Furthermore, the receiver achieves net 220‐Gb s−1 16QAM transmissions in 14 channels from 1527 to 1550 nm with 205‐GHz spacing, and a maximum 240‐Gb s−1 net capacity at a single channel. Such a performance is reached with only standard coherent digital signal processing stacks including all‐linear equalization.

show abstract

The path to 1Tb/s and beyond datacenter interconnect networks: technologies, components, and subsystems

Cited by 10 publications

References 0 publications

InP DHBT Linear Modulator Driver With a 3-Vppd PAM-4 Output Swing at 90 GBaud: From Enhanced Transistor Modeling to Integrated Circuit Design

InP DHBT Linear Modulator Driver With a 3-Vppd PAM-4 Output Swing at 90 GBaud: From Enhanced Transistor Modeling to Integrated Circuit Design

DACON: a reconfigurable application-centric optical network for disaggregated data center infrastructures [Invited]

Spectrally Efficient Integrated Silicon Photonic Phase‐Diverse DD Receiver with Near‐Ideal Phase Response for C‐Band DWDM Transmission

Contact Info

Product

Resources

About