Speaker
Description
Summary
Approximately 40 000 analog readout optical links are being installed in the CMS
Tracker sub-detector for operation in the LHC. The next iteration of the CMS
Tracker will be operated in the Super LHC (SLHC), and will have to cope with
significantly increased data rates due to the tenfold increase in luminosity that
is foreseen. In contrast to the telecom industry where the optical fiber and its
installation drive the cost of a transmission system, it is the cost of the
optoelectronic components that represents a large fraction of the CMS Tracker
electronics budget. Hence, a digital system based on the existing components that
can deliver sufficient performance for SLHC operation, if feasible, could
potentially yield significant financial savings.
Historically, optical transmission systems employ simple binary digital, or on-off
keying (OOK), schemes due to the large bandwidth available in optical fiber. In
contrast, fast transmission through wireless and copper-based communication
channels which are severely band-limited requires modulation techniques that are
more bandwidth-efficient. Research into multi-level signaling for optical systems
is a relatively recent subject of research, prompted by the desire to reach even
higher data rates.
The current CMS Tracker optical readout links are equivalent to a digital system
capable of pulse amplitude modulation (PAM) at 40MS/s with 8bit resolution
(=320Mbps). The excellent noise characteristic (typical SNR ~50dB) means that the
current link is particularly suited to a bandwidth efficient, multi-level digital
modulation scheme. Moreover, the short transmission lengths (~65m) suggest immunity
to attenuation, phase noise and dispersion which are typical performance-limiting
factors in optical transmission systems. It is the bandwidths of the laser driver
chip and receiver amplifier (~100MHz) that ultimately restrict the achievable data
rate.
ADSL is an example of how digital communication theory can be used to fully exploit
a system’s available bandwidth. Copper telephone lines, typically having a 3-dB
bandwidth of ~20kHz, attenuate a signal by as much as 90dB at 1MHz, which is the
outer edge of the frequency band used by ADSL. Using advanced techniques which are
variations of Orthogonal Frequency Division Multiplexing (OFDM) and Quadrature
Amplitude Modulation (QAM), data rates between 2 and 6MBps are accomplished.
Measurements on the Tracker analog links show an attenuation of around 50-60dB at a
frequency of ~700MHz. If similar digital modulation schemes can be employed over
this vastly increased bandwidth and superior Signal to Noise Ratio (SNR), data
rates in the Gbps range could be envisaged.
A simulation using MATLAB Simulink has been developed in order to evaluate the
benefits and drawbacks of various well-documented digital modulation techniques.
The simulation can be used to produce eye diagrams as well as compute BER. The
problem is treated as that of the design of a digital communication system, with
the analog channel (or medium) over which signals are transmitted being the current
Tracker readout link. Extensive testing on the Tracker links allows accurate
modeling of the channel as a frequency-selective filter which adds Additive White
Gaussian Noise (AWGN). Hence the effects of amplitude and phase distortion that
cause Inter Symbol Interference (ISI) are included. In addition to performance
assessment of modulation schemes, the simulation model can be used to identify the
methods of Forward Error Correction (FEC) that are appropriate for this particular
system. Moreover, source encoding (i.e. at the front end) in the form of data
compression or optimal digitization of the analog detector signals can be easily
explored.
The results of the study will demonstrate the feasibility of maximizing the
bandwidth utilization of the current CMS Tracker optical links to achieve a Gbps
link, and guide in the choice of method required. The application of advanced
communication theory in the context of a HEP experiment is novel and would, in
practice, present serious technological challenges. Nevertheless, the knowledge
gained in this project could benefit future experiments by providing insight into
fundamental engineering concepts never implemented before in HEP instrumentation.
