Ali Ghiasi

Higher-order modulation for client optics

Higher-order modulation is being studied for future 100G, 400G, and faster client optics. Amplitude, phase, and multi-subcarrier modulation approaches, used separately and in combination with each other, are being considered. Increased penalties due to higher-order modulation greater sensitivity to link impairments require use of DSP algorithms implemented in advanced CMOS nodes to close the link budget. DSP also enables programmability to reconfigure the optics for different applications and link conditions.

Is there a need for on-chip photonic integration for large data warehouse switches

Over the past 18-months VCSEL based optical engines have been integrated into package of large-scale HPC routers, moderate size Ethernet switches, and even FPGAs. Competing solutions based on Silicon Photonics (SiP) are emerging and targeting similar application space but with better integration path through the use of TSV (Through Silicon Via) stack dies. Integrating either VCSEL or SiP based optical engines into complex IC package that operate at high temperatures and require high reliability is not trivial, and one should ask what is the technical or the economic advantage before embarking on such a drastic architectural change. We are currently investigating the extend to which scenarios on-chip photonic integration into large-scale data warehouse switches have clear power and/or bandwidth advantages over using traditional electrical I/O.

Next-generation 10 GBaud module based on emerging SFP+ with host-based EDC [Topics in Optical Communications]

There has been an ongoing trend to require transceivers for use in datacom and telecom switches to be small and have low power dissipation to enable large port count switches. At speeds between 1 and 4 GBaud the SFP form factor is by far the most commonly used. Up to now 10 GBaud transceivers have used larger devices with larger power dissipations. This article describes the SFP+ module being specified by the SFF Committee (SFF 8431) that will enable the same port densities as the SFP module. The adaptive equalizers and high-speed transmission channels required in the switches to make the SFP+ module work successfully are also described

Experimental results of EDC based receivers for 2400 ps/nm at 10.7 Gb/s for emerging telecom standards

Extensive experimental work with electronic dispersion compensation, a critical new technology for Metro 10G networks, was carried out by multiple companies. The results supporting 2400 ps/nm link are presented and form the bases for emerging Telecom application codes.

Feasibility of unretimed 100 GbE based on 4×25.78 GBd

We have demonstrated the feasibility of unretimed 100 GbE based on 4×25.78 GBd signaling, using a 40 nm CMOS SerDes interfaced to a 100 mm host PCB channel that includes a z-pluggable connector connected to a 100Gbase-LR4 ROSA.

Enabling 850nm VCSELs for 100GbE unretimed applications

We have investigated feasibility and limitations of 850nm VCSELs for 100GbE unretimed applications based on 4×25.78GBd signaling. We have shown traditional FFE/DFE based equalizer could overcome optical as well as electrical channel impairments.

Enabling 850nm VCSELs for 100GbE applications

We have investigated feasibility and limitations of 850nm VCSELs for 100GbE applications based on 4×25.78GBd signaling. We showed traditional FFE/DFE based equalizer could overcome optical as well as electrical impairments.

Benefits of EDC and linear receivers for short reach 40/100GE

An ongoing trend in datacom switches is to require optical transceivers to be smaller and have lower power dissipation in order to enable greater port densities. The transition from 1 GE to 10 GE was unusually slow for the Ethernet market due in part to the high cost. high power dissipation, and form factor of the optical module. The industry is now overcoming these hurdles with the transition to SFP + modules, which will enable more traditional Ethernet growth. Upcoming 40 GE and 100 GE are multilane protocols, presenting even greater challenges than 10 GE laced at its inception, potentially slowing their adoption. Leveraging SFP+ class technology such as simple EDC capabilities can lower cost and reduce module form factors, thus accelerating the transition to 40/100 GE interlace deployment.

InfiniBand—The Cluster Interconnect

Publisher Summary This chapter describes InfiniBandTM (IB), which is an optical link interface. The InfiniBand group was formed to facilitate development of a uniform interconnect from back plane to data center links. The InfiniBand link provides an interoperable interface with raw bandwidth of 250 MBytes/s to 12 Gbyte. The InfiniBand link layer provides a connection between two InfiniBand protocol aware ports. Each physical link may have 1, 4, 8, or 12 physical lanes. A physical link may be copper cable, copper back plane, or fiber-optics. InfiniBand allows the connection of two protocol aware devices with different width. During link negotiation, the link width common denominator is established between two protocol aware devices, possibly a single lane wide. Today InfiniBand is well established in the cluster interconnect, but is not deployed as widely as envisioned due to the market meltdown of 2001, introduction of PCI Express, and 10- Gigabit Ethernet.

8.5 gigabit and 10 gigabit SFP+ module enabling next-generation high density switches [SFP+]

Discusses the SFP+ project.