Optically transparent multiple access networks employing incoherent spectral codes

This Ph.D. thesis is divided into 7 chapters to provide the reader an overview of the main results achieved in di®erent sub-topics of the study towards optically transparent multiple access networks employing incoherent spectral codes taking into account wireless transmission aspects. The work contains Layer 1 (photonic integration, ¯ber-optic transmission), Layer 2 (medium access control) and Layer 3 (packet switching, transmission scheduling) related material with a strong focus on the optical physical layer, Layer 1. A brief summary is given of the contents per chapter. Chapter 1 "Introduction" places the research done in this thesis within the context of current networking trends and emerging communication scenarios. On the one hand, optical technologies are inevitable in order to ful¯ll the increasing demand for bandwidth and, on the other hand, mobility through a wireless con- nection remains to be an important asset of the networking experience as a whole. An access tier is proposed whereby the focus is on overall reduction of costs and complexity through an e±cient sharing of resources, simpli¯ed network nodes and centralized network management to transparently carry data of wired and wireless networks in various °avors. As a result, the architecture is based on power-splitting passive optical networks (PONs), optical code division multiple access (OCDMA) and radio-over-¯ber (RoF). In OCDMA, a unique and (pseudo-)orthogonal optical code (OC) signs each data bit transmitted by a user which allows the carrier to be asynchronously shared with other users on the network. RoF techniques en- able to perform the access control and signal processing of a wireless system at the central node, and to deliver transparently the radio signals to simpli¯ed radio access points (RAPs) via optical ¯ber. An architecture based on PON, OCDMA and RoF matches with the design constraints. A large portion of the chapter is devoted to OCDMA systems and their per- formance (issues). Out of the many available techniques, incoherent spectral am- plitude encoded OCDMA (SAE OCDMA) is depicted as the preferred OCDMA iv SUMMARY technique in this thesis. SAE OCDMA o®ers a cost-e®ective system by means of integrateable Mach-Zehnder interferometer (MZI) based en/decoders (E/Ds) and incoherent broad spectral sources. Other OCDMA systems employ short pulse sources which complicate the setup but o®er a higher performance. SAE OCDMA is chosen based on its low complexity and potentially low system costs. It is shown that the SAE OCDMA system su®ers from multiple user interference (MUI), op- tical intensity noise and chromatic dispersion. Note that MUI a®ects all OCDMA techniques. Finally, the remaining of the chapter presents the developments in (emerging) user-centric or personal networking as a motivation behind the work done in this thesis. Chapter 2 "Transparent optical access in personal networks" illus- trates the eminent role of optical transparent networking in personal networks (PNs). A PN is de¯ned by the personal area network (PAN) of a user and the other, remote personal devices the PAN is connected with. A PAN emerges when the network-aware devices carried by a user interconnect via a short-range wireless interface such as Bluetooth and constitutes a small, private and literally mobile network. A PN then extends the local scope of a PAN to a global scope but with- out modi¯cation of its personalized character. It is shown how an optical access network solves many issues with respect to connectivity and mobility manage- ment in PNs. Additionally, a concrete PN user-scenario is given which deduces requirements for such an architecture. Chapter 3 "Incoherent spectral amplitude OCDMA" focusses on the SAE OCDMA technique which was introduced in chapter 1. Two major topics are treated in this chapter namely the modular construction of E/Ds and the orthogonality of the code set. A novel parallel spectral (or tree) E/D is presented and its performance is analyzed in detail. Compared with a well-known serial spectral (or cascaded) E/D, the tree E/D only introduces a power penalty due to the parallel en/decoding operation. The performance of the SAE optical code (OC) set is analyzed in terms of crosstalk. As the code set grows in size, full orthogonality is lost and crosstalk (MUI) becomes an issue. Furthermore it is shown that reasonable fabrication inaccuracies in the building blocks of the E/Ds can be tolerated by the code set. At the end of the chapter, enhancing the spectral e±ciency is brie°y considered via channel interleaving in an SAE OCDMA system. Chapter 4 "Photonic integration of multi-stage en/decoders" de- scribes the photonic integration of multi-stage cascaded and tree E/D structures in two di®erent kinds of material systems namely InP/InGaAsP and Si3N4/SiO2. The InP/InGaAsP material system enables electro-optic phase shifting and the monolithic integration of passive and active components such as a semiconduc- tor optical ampli¯er for a scalable system-on-a-photonic-chip design. However, the typical propagation losses are in the order of dB per cm and a complicated fabrica- tion process is used. Si3N4/SiO2 provides lower propagation losses in the order of tenths of dB per cm and is compatible with standard CMOS production processes. However, thermo-optic phase shifters are used which consume more energy and a similar design requires a larger chip area in Si3N4/SiO2 than in InP/InGaAsP. In this chapter, the full trajectory is shown of designing, processing and characteriz- ing multiple en/decoder con¯gurations. The measured optical characteristics of the realized photonic chips have an excellent match with the designed and modelled properties. The SAE OCDMA en/decoders manufactured in the InP/InGaAsP material system have the highest potential in terms of low capital and opera- tional expenses when compared with SAE OCDMA en/decoders manufactured in the Si3N4/SiO2 material system considering losses, fabrication process, energy consumption, scalability and usage of chip area although other factors such as packaging and polarization dependency also have to be carefully considered. Chapter 5 "SAE OCDMA transmission performance" analyses the transmission performance of the SAE OCDMA technique on a PON via simu- lations with the cascaded and tree E/Ds at the optical networking unit (ONU) and central o±ce (CO), respectively. The simulated results of up and downstream transmission are presented considering a user bit rate of 1 Gbps. It is shown that MUI and intensity noise severely degrade the performance. Only three users can transmit error-free by employing forward error correction (FEC) techniques which add complexity and a transmission overhead to the system. It is con¯rmed by simulation that increasing the received power does not improve the system's MUI performance. At the end of the chapter, downstream system experiments are dis- cussed and results are shown on measured crosstalk values, the contrast ratios between a code match/mismatch, and on a successful error-free transmission ex- periment for a single user at a Gigabit Ethernet (GbE) speed. Chapter 6 "Code-based transparent access tier" introduces code-based all-optical communication between ONUs in a sub-net of PONs such that an op- tical transparent network is realized. However, collisions may occur between data transmitted from di®erent ONUs to a single destination therefore network con- tention resolution methods are required. In order to preserve the transparency in the network, the network management functionalities are implemented in the opti- cal domain. Optical code-sense multiple access / collision detection (OCSMA/CD) is proposed to resolve contention when multiple ONUs share a PON. Optical code- labelled packet switching is considered to resolve contention when two ONUs in the same sub-net do not share a PON. The implementations are discussed for a generic case allowing any kind of OCDMA system. OCSMA/CD requires a re- °ective PON which introduces a limitation in the maximum achievable splitting ratio. Besides, the minimum allowable packet size in the network restricts the ¯ber length between the ONU and the re°ective coupler. The packet switch adds signi¯- cant complexity to the CO which becomes more profound when a larger number of PONs and ONUs have to be accommodated. Therefore depending on the network parameters, the implementation of optical network contention resolution has to be carefully considered. The code-sensing mechanism may also be used for continuous monitoring of network activity. Based on that information, it is shown that the MUI e®ect can be mitigated via the avoidance of interference. This has been eval- uated via numerical experiments for the SAE OCDMA system in case of di®erent packet size distributions. vi SUMMARY Chapter 7 "Microwave service delivery in FTTPAN" presents explo- rative research done towards broadband service delivery independent of the user's location employing an optical ¯ber network architecture. Three main categories of RoF systems are considered in detail and the harmonics generation technique denoted as optical frequency multiplication (OFM) is found to have the optimal match with an SAE OCDMA system. The OFM/SAE-OCDMA combination is analyzed on a PON and, as a result, it is concluded that the RoF technique can only be used for the distribution of radio frequency (RF) signals on PON while the OCDMA system should be used for the transmission of RAP-speci¯c data. OCDMA inherently is a digital transmission technique therefore analogue mod- ulation formats are not supported. Therefore a system is proposed in which the analogue data is remotely "constructed" at the ONU side, that is, the digital and analogue (RF) signals are transmitted separately via OCDMA and OFM after which remote modulation and mixing is performed. The proof of concept is shown by a simulated hybrid OFM/SAE-OCDMA system which successfully transmits quadrature phase shift keying (QPSK) data at a symbol rate of 500 MSymbols/s. Polarization multiplexing (PolMUX) is employed to send the I and Q channels of the QPSK signal by only using a single code on the network. Additionally, it is con¯rmed that a lower operational bit rate improves the system's MUI perfor- mance.