MEMS-tunable 10 Gb/s APD receiver for
broadcast and select CATV networks
Jill D. Berger, Doug Anthon, Joe Drake, and Fedor Ilkov
Iolon, Inc., 1870 Lundy Ave., San Jose, California 95131
jberger@iolon.com
Jason Shreeram and Carlos Verdonck
Scientific-Atlanta, Inc., 5030 Sugarloaf Parkway, Lawrenceville, GA 30044
jason.shreeram@sci-atl.com
Abstract: A 10 Gb/s receiver combining a MEMS-tunable optical filter and avalanche photodiode in a
common butterfly provides tuning over 5 THz with –26 dBm receiver sensitivity and ± 5 GHz
wavelength control, to enable dynamic provisioning for broadcast and select CATV networks.
© 2004 Optical Society of America
OCIS Codes: (060.2380) Fiber optics sources and detectors; (060.4250) Networks
1. Network architecture
The term convergence is used in the cable television (CATV) industry to describe the deployment and management of
multiple signals in different formats over a single network. CATV traffic consists of a bi-directional mixture of analog
video signals and mixed-protocol digital signals, with digital video, high speed data, Gigabit Ethernet, TDM voice and
DVB/ITU/ASI all being used in various systems. These signals are managed so they can be efficiently transmitted through
the network and connected to the end user via a single coaxial cable or optical fiber. Convergence can take place in either
the electrical, optical or system domain, with the network architecture being chosen to optimize this process. Convergence
in the electrical domain is carried out using a variety of IEEE and ITU standards, such as SONET Layer 1, Resilient Packet
Ring, and MPLS. Convergence in the optical domain requires wavelength management and optical performance
monitoring, not only for conventional base-band digital signals but also for sub-carrier modulated analog video and digital
QAM signals, in both the forward and reverse directions. In the system domain an efficient transition must be made
between the requirements of the different optical signals and the requirements of coaxial cable. The high capacity optical
fiber links need to be managed in a way that is compatible with existing multi-vendor network equipment and RF signal
requirements. Efficient convergence at all levels is required to ensure that a digital signal transmitted over a fiber as
Gigabit Ethernet in one part of the system can ultimately be transmitted to an end user as a sub-carrier multiplexed 256
QAM electrical signal.
One important growth area for CATV networks is that of video on demand (VOD), which requires efficient
distribution of a large number of data streams through the network. The problem, in this case, is to provide an efficient
distribution network that has sufficient flexibility to accommodate unpredictable traffic fluctuations without requiring
excessive initial investment. One method for achieving this goal involves a tunable multi-wavelength bi-directional digital
ring at 10 Gb/s in a broadcast and select architecture with tunable lasers at each head end and tunable receivers at each
network node [1]. Figure 1 shows the basic node in this architecture, where a combination of multi-band filters and
broadband tap couplers are used to provide a set of fixed wavelength channels as well as a large number of wavelengths
that can be accessed in either fiber direction by the tunable receiver used for the VOD signals. The purpose of the fixed
wavelength channels is to accommodate legacy services presently transported on individual fibers. The availability of these
channels is important for integrating new and existing services into the network and is an important benefit of increased
flexibility in the optical layer.
As initially deployed, the number of tunable lasers at the head end will be determined by anticipated traffic levels,
with a few additional lasers being necessary to provide redundancy. The number of lasers can be smaller than the number of
nodes, as the less-used nodes can share a single transmitter. Each node will have two tunable receivers, allowing for both
redundancy and traffic growth. Each laser broadcasts in both directions around the ring, so the 1×2 switch at the node
provides redundancy in the event of a fiber failure. Using tunable devices at both the laser and receiver enables the network
to be provisioned for dynamic optical switching and bandwidth on demand [2]. The principle advantage of the broadcast-
and-select architecture is the ability to start with a bandwidth model for today’s network and migrate to a future higher
bandwidth model through software. In order to efficiently deploy new VOD services in existing networks as well as meet
future traffic demands, the network must be capable of dynamic asset deployment. By placing a broadband splitter at the
input of a VOD receiving hub, sufficient optical level can be designed into the network to accommodate the everything-on-
demand network model and easily adapt the network to transport up to 10 times the capacity of the initial design.