Performance Enhancement in Copper Twisted Pair Cable Communications
Date
Authors
Advisors
Journal Title
Journal ISSN
ISSN
DOI
Volume Title
Publisher
Type
Peer reviewed
Abstract
The thesis focuses on the area of copper twisted pair based wireline communications. As one of the most widely deployed communication media, the copper twisted pair cable plays an important role in the communication network cabling infrastructure. This thesis looks to exploit diversity to improve twisted pair channels for data communications in two common application areas, namely Ethernet over Twisted Paris and digital subscriber line over twisted pair based telephone network.
The first part of the thesis addresses new approaches to next generation Ethernet over twisted pair cable. The coming challenge for Ethernet over twisted pair cable is to realise a higher data rate beyond the 25/40GBASE-T standard, in relatively short reach scenarios. The straight-forward approaches, such as improving cable quality and extending frequency bandwidth, are unlikely to provide significant improvement in terms of data rate. However, other system diversities, such as spectrum utilization are yet to be fully exploited, so as to meet the desired data rate performance. The current balanced transmission over the structured twisted pair cable and its parallel single-in-single-out channel model is revisited and formulated as a full-duplex multiple-in-multiple-out (MIMO) channel model.
With a common ground (provided by the cable shield), the balanced transmission is converted into unbalanced transmission, by replacing the differential-mode excitation with single-ended excitation. In this way, MIMO adoption may offer spectrum utilization advantages due to the doubled number of the channels. The S-parameters of the proposed MIMO channel model is obtained through the full wave electromagnetic simulation of a short CAT7A cable. The channel models are constructed from the resulting S-parameters, also the corresponding theoretical capacity is evaluated by exploiting different diversity scenarios. With higher spectrum efficiency, the orthogonal-frequency-division-multiplexing (OFDM) modulation can significantly improve the theoretical capacity compared with single-carrier modulation, where the channel frequency selectivity is aided. The MIMO can further enhance the capacity by minimising the impact of the crosstalk. When the crosstalk is properly handled under the unbalanced transmission, this thesis shows that the theoretical capacity of the EoTP cable can reach nearly 200GBit/s. In order to further extend the bandwidth capability of twisted pair cables, Phantom Mode transmission is studied, aiming at creating more channels under balanced transmission operation. The second part of the thesis focuses on the research of advanced scheduling algorithms for VDSL2 QoS enhancement. For VDSL2 broadband access networks, multi-user optimisation techniques have been developed, so as to improve the basic data rate performance. Spectrum balancing improves the network performance by optimising users transmit power spectra as the resource allocation, to mitigate the impact from the crosstalk. Aiming at enhancing the performance for the upstream VDSL2 service, where the users QoS demand is not known by all other users, a set of autonomous spectrum balancing algorithms is proposed. These optimise users transmit power spectra locally with only direct channel state information. To prevent selfish behaviour, the concept of a virtual user is introduced to represent the impact on both crosstalk interference and queueing status of other users. Moreover, novel algorithms are developed to determine the parameters and the weight of the virtual user. Another type of resource allocation in the VDSL2 network is crosstalk cancellation by centralised signal coordination. The history of the data queue is considered as a time series, on which different smooth filter characteristics are investigated in order to investigate further performance improvement. The use of filter techniques accounts for both the instantaneous queue length and also the previous data to determine the most efficient dynamic resource allocation. With the help of this smoothed dynamic resource allocation, the network will benefit from both reduced signalling communication and improved delay performance.The proposed algorithms are verified by numerical experiments.