The continuing growth of microelectronics in speed, scale, and complexity has led to a looming bandwidth bottleneck for traditional electronic interconnects. This has precipitated the penetration of optical interconnects to smaller, more localized scales, in such applications as data centers, supercomputers, and access networks. For this next generation of optical interconnects, the silicon photonic platform has received wide attention for its ability to manifest, more economical, high-performance photonics. The high index contrast and CMOS compatibility of the silicon platform give the potential to intimately integrate small footprint, power-efficient, high-bandwidth photonic interconnects with existing high-performance CMOS microelectronics. Within the silicon photonic platform, traditional photonic elements can be manifested with smaller footprint and higher energy-efficiency. Additionally, the high index contrast allows the successful implementation of silicon microring-based devices, which push the limits on achievable footprint and energy-efficiency metrics. While laboratory demonstrations have testified to their capabilities as powerful modulators, switches, and filters, the commercial implementation of microring-based devices is impeded by their susceptibility to fabrication tolerances and their inherent temperature sensitivity. This work develops and demonstrates methods to resolve the aforementioned sensitivities of microring-based devices. Specifically, the use of integrated heaters to thermally tune and lock microring resonators to laser wavelengths, and the underlying control systems to enable such functionality. The first developed method utilizes power monitoring to show the successful thermal stabilization of a microring modulator under conditions that would normally render it inoperational. In a later demonstration, the photodetector used for power monitoring is co-integrated with the microring modulator, again demonstrating thermal stabilization of a microring modulator and validating the use of defect-enhanced silicon photodiodes for on-chip control systems. Secondly, a generalized method is developed that uses dithering signals to generate anti-symmetric error signals for use in stabilizing microring resonators. A control system utilizing a dithering signal is shown to successfully wavelength lock and thermally stabilize a microring resonator. Characterizations are performed on the robustness and speed of the wavelength locking process when using dithering signals. An FPGA implementation of the control system is used to scale to a WDM microring demultiplexer, demonstrating the simultaneous wavelength locking of multiple microring resonators. Additionally, the dithering technique is adopted to create control systems for microring-based switches, which have traditionally posed a challenging problem due to their multi-state configurations. The aforementioned control systems are rigorously tested for applications with high speed data and analyzed for power efficiency and scalability to show that they can successfully scale to commercial implementations and be the enabling factor in the commercial deployment of microring-based devices.

With the rapid development of energy harvesting technologies, a new paradigm of wireless communications that employs energy harvesting transmitters has become a reality. The renewable energy source enables the flexible deployment of the transmitters and prolongs their lifetimes. To make the best use of the harvested energy, many challenging research issues arise from the new paradigm of communications. In particular, optimal resource (energy, bandwidth, etc.) allocation is key to the design of an efficient wireless system powered by renewable energy sources. In this thesis, we focus on several resource allocation problems for energy harvesting communications, including the energy allocation for a single energy harvesting transmitter, and the joint energy and spectral resource allocation for energy harvesting networks. More specifically, the resource allocation problems discussed in this thesis are summarized as follows. We solve the problem of designing an affordable optimal energy allocation strategy for the system of energy harvesting active networked tags (EnHANTs), that is adapted to the identification request and the energy harvesting dynamic. We formulate a Markov decision process (MDP) problem to optimize the overall system performance which takes into consideration of both the system activity-time and the communication reliability. To solve the problem, both a static exhaustive search method and a modified policy iteration algorithm are employed to obtain the optimal energy allocation policy. We develop an energy allocation algorithm to maximize the achievable rate for an access-controlled energy harvesting transmitter based on causal observations of the channel fading states. We formulate the stochastic optimization problem as a Markov decision process (MDP) with continuous states and define an approximate value function based on a piecewise linear fit in terms of the battery state. We show that with the approximate value function, the update in each iteration consists of a group of convex problems with a continuous parameter and we derive the optimal solution to these convex problems in closed-form. Specifically, the computational complexity of the proposed algorithm is significantly lower than that of the standard discrete MDP method. We propose an efficient iterative algorithm to obtain the optimal energy-bandwidth allocation for multiple flat-fading point-to-point channels, maximizing the weighted sum-rate given the predictions of the energy and channel state. For the special case that each transmitter only communicates with one receiver and the objective is to maximize the total throughput, we develop efficient algorithms for optimally solving the subproblems involved in the iterative algorithm. Moreover, a heuristic algorithm is also proposed for energy-bandwidth allocation based on the causal energy and channel observations. We consider the energy-bandwidth allocation problem in multiple orthogonal and non-orthogonal flat-fading broadcast channels to maximize the weighted sum-rate given the predictions of energy and channel states. To efficiently obtain the optimal allocation, we extend the iterative algorithm originally proposed for multiple flat-fading point-to-point channels and further develop the optimal algorithms to solve the corresponding subproblems. For the orthogonal broadcast channel, the proportionally-fair (PF) throughput maximization problem is formulated and we derive the equivalence conditions such that the optimal solution can be obtained by solving a weighted throughput maximization problem. The algorithm to obtain the proper weights is also proposed. We consider the energy-subchannel allocation problem for energy harvesting networks in frequency-selective fading channels. We first assume that the harvested energy and subchannel gains can be predicted and propose an algorithm to efficiently obtain the energy-subchannel allocations for all links over the scheduling period based on controlled water-filling. The proposed algorithm is shown to be asymptotically optimal when the bandwidth of the subchannel goes to zero. A causal algorithm is also proposed based on the Q-learning technique that makes use of the statistics of the energy harvesting and channel fading processes.

Silicon has numerous benefits as a photonic integrated circuit platform, including optical transparency from 1.1 µm to greater than 5 µm, tight optical confinement due to its high index of refraction, high third order non-linearity, and lack of two photon absorption at wavelengths above 2.2 µm. Additionally, silicon photonics has the added benefit of decades of fabrication knowledge from the CMOS industry. Despite these advantages, an enormous challenge exists in two areas, optical sources for silicon photonic integrated circuits, and on the other end, optical detectors for silicon photonic integrated circuits. The same bandgap energy that leads to the optical transparency at telecom and mid-infrared wavelengths, limits both generation and detection in this same regime. This dissertation focuses on the detection problem, exploring the use of defect-mediated sub-bandgap optical absorption in ion-implanted silicon nano-wire waveguides. Section I of this dissertation focuses on fabrication and the ion-implantation process, including a primer on Shockley-Read-Hall theory and its application to defect-mediated sub-bandgap optical absorption. Section II examines the devices for use at telecom wavelengths. In Chapter 4, the fabrication and characterization of metal-semiconductor-metal ion-implanted silicon nano-wire waveguide photodiodes is examined. These devices require minimal fabrication, are compatible with standard CMOS fabrication processes, and exhibited responsivities as high as 0.51 A/W and frequency responses greater than 2.6 GHz. With improved fabrication tolerances, frequency responses of greater than 10 GHz are expected. Chapter 5 examined these ion-implanted photodiodes in a p-i-n configuration as a high speed data interconnect, demonstrating error free operation at 10 Gbs with expected sensitivities approaching that of Ge detectors. Section III extends the above research to longer wavelengths, starting with data reception at 1.9 µm in Chapter 6, exhibiting an approximate 5 dB penalty in sensitivity compared to the same diodes at 1.55 µm, at a data rate of 1 Gbs, limited by RC due to the 2 mm length of the device. Chapter 7 goes even further, characterizing Si+ implanted silicon nano-wire waveguides for operation between 2.2 µm and 2.35 µm. These devices showed responsivities as high as 9.9 mA/W, with internal quantum efficiencies approaching 5%. Chapter 8 concludes with the characterization of Zn+ implanted silicon nano-wire waveguides operating in the same wavelength regime, exhibiting higher overall responsivity, albeit at a much higher reverse bias. These long wavelength devices open up new areas of research for silicon photonics, allowing for CMOS compatible detectors operating into the mid-infrared region, useful for chemical sensing, free-space communications, and medical imaging.

Today, the Internet traffic is growing at a near exponential rate, driven predominately by data center-based applications and Internet-of-Things services. This fast-paced growth in Internet traffic calls into question the ability of the existing optical network infrastructure to support this continued growth. The overall optical networking equipment efficiency has not been able to keep up with the traffic growth, creating a energy gap that makes energy and cost expenditures scale linearly with the traffic growth. The implication of this energy gap is that it is infeasible to continue using existing networking equipment to meet the growing bandwidth demand. A redesign of the optical networking platform is needed. The focus of this dissertation is on the design and implementation of energy efficient, cross-layer enabled, dynamic optical networking platforms, which is a promising approach to address the exponentially growing Internet bandwidth demand. Chapter 1 explains the motivation for this work by detailing the huge Internet traffic growth and the unsustainable energy growth of today's networking equipment. Chapter 2 describes the challenges and objectives of enabling agile, dynamic optical networking platforms and the vision of the Center for Integrated Access Networks (CIAN) to realize these objectives; the research objectives of this dissertation and the large body of related work in this field is also summarized. Chapter 3 details the design and implementation of dynamic networking platforms that support wavelength switching granularity. The main contribution of this work involves the experimental validation of deep cross-layer communication across the optical performance monitoring (OPM), data, and control planes. The first experiment shows QoS-aware video streaming over a metro-scale test-bed through optical power monitoring of the transmission wavelength and cross-layer feedback control of the power level. The second experiment extends the performance monitoring capabilities to include real-time monitoring of OSNR and polarization mode dispersion (PMD) to enable dynamic wavelength switching and selective restoration. Chapter 4 explains the author's contributions in designing dynamic networking at the sub-wavelength switching granularity, which can provide greater network efficiency due to its finer granularity. To support dynamic switching, regeneration, adding/dropping, and control decisions on each individual packet, the cross-layer en- abled node architecture is enhanced with a FPGA controller that brings much more precise timing and control to the switching, OPM, and control planes. Furthermore, QoS-aware packet protection and dynamic switching, dropping, and regeneration functionalities were experimentally demonstrated in a multi-node network. Chapter 5 describes a technique to perform optical grooming, a process of optically combining multiple incoming data streams into a single data stream, which can simultaneously achieve greater bandwidth utilization and increased spectral efficiency. In addition, an experimental demonstration highlighting a fully functioning multi-node, agile optical networking platform is detailed. Finally, a summary and discussion of future work is provided in Chapter 6. The future of the Internet is very exciting, filled with not-yet-invented applications and services driven by cloud computing and Internet-of-Things. The author is cautiously optimistic that agile, dynamically reconfigurable optical networking is the solution to realizing this future.

The accelerated growth in performance of microprocessors and the emergence of chip multiprocessors, which are now widely leveraged in current data centers and high-performance computing (HPC) systems, have motivated the need for developing novel interconnection networks solutions to meet the growing need for data transmissions across all levels of the infrastructure. This work posits that, given the unique characteristics of optics---advantages and limitations---purpose-driven systems-level designs are necessary in order to harness the tremendous performance and efficiency opportunities that can be enabled by photonic interconnects. First, an enhanced optically connected network architecture is presented featuring advanced photonic functionalities to support a wider class of bandwidth-intensive traffic patterns characteristic of cloud computing systems. This proposed architectural framework can enable a rich set of photonic resources to be allocated on-demand to optimize communications between various applications within the data center. A prototype of the proposed optical network architecture is constructed and a demonstration of two unique functionalities, serving to validate the physical layer feasibility of the system, is presented. An instantiation of this architectural framework is presented that enables physical layer data duplication in order to more effectively support reliable group data delivery in the data center. Compared to the conventional solutions that duplicate data in the network or application layer, this architecture achieves efficient data transmission over the ultra-fast, loss-free, energy-efficient and low cost optical paths, with simplified flow control, congestion control, and group membership management. Both an end-to-end hardware experiment and large-scale simulations were carried out to evaluate the efficacy of the design. Next, the challenges associated with interfacing to photonically-switched networks are explored. In particular, various interface designs aimed at addressing the unique challenges imposed by optical-packet switched networks are proposed and evaluated. First, an overview of the data vortex network optical packet switch architecture is given. A high-speed optical packet formatter and interface is then presented along with the results of end-to-end data exchanges across the interface connected to a data vortex network. Finally, the design of a low-power all-optical interface alternative is validated with an end-to-end demonstration. Finally, various unique photonic switching node designs are introduced for a variety of applications|a nanosecond-scale bidirectional 2 x —2 switch to construct efficient optical fat-tree architectures, a 4 x —4 switch capable of operating as both a nanosecond-scale optical packet switch and as an optical circuit switch, and a non-blocking 4 x —4 switch designed for constructing on-chip photonic integrated networks.

In recent years the proliferation of wireless devices such as tablets and smartphones has resulted in an unprecedented crowding of the radio spectrum around the world. The high density of radio signals being transmitted at any one time has necessitated the use of high-performance radio-frequency (RF) filters prior to any receiver signal path in order to protect the device against interference. State-of-the-art filter technologies based on piezoelectric resonators provide good rejection of interfering signals, but do not scale well to cover the large range of frequencies currently allocated for cellular communications. This thesis presents measurements and analysis of a new active resonator technology, known as a graphene resonant channel transistor (G-RCT), that has the potential to be used in micron-scale RF filters that are capable of covering these larger bandwidths. A compact model for G-RCTs is developed that accurately replicates the AC, DC, and frequency tuning characteristics of the device, enabling the design and simulation of hybrid electromechanical circuits. The device noise is also modeled, and analytical expressions for the noise figure of circuits using G-RCTs are derived. Finally, expressions for third-order intermodulation distortion are derived and validated with measurements.

Growing demands in heterogeneous utility computing systems in future cloud and high performance computing systems are driving the development of processor-hardware accelerator interconnects with greater performance, flexibility, and dynamism. Recent innovations in the field of utility computing have led to an emergence in the use of heterogeneous compute elements. By leveraging the computing advantages of hardware accelerators alongside typical general purpose processors, performance efficiency can be maximized. The network linking these compute nodes is increasingly becoming the bottleneck in these architectures, limiting the hardware accelerators to be restricted to localized computing. A high-bandwidth, agile interconnect is an imperative enabler for hardware accelerator delocalization in heterogeneous utility computing. A redesign of these systems' interconnect and architecture will be essential to establishing high-bandwidth, low-latency, efficient, and dynamic heterogeneous systems that can meet the challenges of next-generation utility computing. By leveraging an optics-based approach, this dissertation presents the design and implementation of optically-connected hardware accelerators (OCHA) that exploit the distance-independent energy dissipation and bandwidth density of photonic transceivers, in combination with the flexibility, efficiency and data parallelization offered by optical networks. By replacing the electronic buses with an optical interconnection network, architectures that delocalize hardware accelerators can be created that are otherwise infeasible. With delocalized optically-connected hardware accelerator nodes accessible by processors at run time, the system can alleviate the network latency issues plague current heterogeneous systems. Accelerators that would otherwise sit idle, waiting for it's master CPU to feed it data, can instead operate at high utilization rates, leading to dramatic improvements in overall system performance. This work presents a prototype optically-connect hardware accelerator module and custom optical-network-aware, dynamic hardware accelerator allocator that communicate transparently and optically across an optical interconnection network. The hardware accelerators and processor are optimized to enable hardware acceleration across an optical network using fast packet-switching. The versatility of the optical network enables additional performance benefits including optical multicasting to exploit the data parallelism found in many accelerated data sets. The integration of hardware acceleration, heterogeneous computing, and optics constitutes a critical step for both computing and optics. The massive data parallelism, application dependent-location and function, as well as network latency, and bandwidth limitations facing networks today complement well with the strength of optical communications-based systems. Moreover, ongoing efforts focusing on development of low-cost optical components and subsystems that are suitable for computing environment may benefit from the high-volume heterogeneous computing market. This work, therefore, takes the first steps in merging the areas of hardware acceleration and optics by developing architectures, protocols, and systems to interface with the two technologies and demonstrating areas of potential benefits and areas for future work. Next-generation heterogeneous utility computing systems will indubitably benefit from the use of efficient, flexible and high-performance optically connect hardware acceleration.

In the near future, objects equipped with heterogeneous devices such as sensors, actuators, and tags, will be able to interact with each other and cooperate to achieve common goals. These networks are termed the Internet of Things (IoT) and have applications in healthcare, smart buildings, assisted living, manufacturing, supply chain management, and intelligent transportation. The IoT vision is enabled by ubiquitous wireless communications and there are numerous resource allocation challenges to efficiently connect each device to the network. In this thesis, we study wireless resource allocation problems that arise in the IoT, namely in the areas of the energy harvesting tags, termed the Internet of Tags (IoTags), and in cellular networks (mobile and cognitive). First, we present our experience designing and developing Energy Harvesting Active Networked Tags (EnHANTs). The prototypes harvest indoor light energy using custom organic solar cells, communicate and form multihop networks using ultra-low-power Ultra- Wideband Impulse Radio (UWB-IR) transceivers, and dynamically adapt their communications and networking patterns to the energy harvesting and battery states. Using our custom designed small scale testbed, we evaluate energy-adaptive networking algorithms spanning the protocol stack (link, network, and flow control). Throughout the evaluation of experiments, we highlight numerous phenomena which are typically difficult to capture in simulations and nearly impossible to model in analytical work. We believe that these lessons would be useful for the designers of many different types of energy harvesters and energy harvesting adaptive networks. Based on the lessons learned from EnHANTs, we present Power Aware Neighbor Discovery Asynchronously (Panda), a Neighbor Discovery (ND) protocol optimized for networks of energy harvesting nodes. To enable object tracking and monitoring applications for IoTags, Panda is designed to efficiently identify nodes which are within wireless communication range of one another. By accounting for numerous hardware constraints which are typically ignored (i.e., energy costs for transmission/reception, and transceiver state switching times/costs), we formulate a power budget to guarantee perpetual ND. Finally, via testbed evaluation utilizing Commercial Off-The-Shelf (COTS) energy harvesting nodes, we demonstrate experimentally that Panda outperforms existing protocols by a factor of 2-3x. We then consider Proportional Fair (PF) cellular scheduling algorithms for mobile users, These users experience slow-fading wireless channels while traversing roads, train tracks, bus routes, etc. We leverage the predicable mobility on these routes and present the Predictive Finite-horizon PF Scheduling ((PF)2S) Framework. We collect extensive channel measurement results from a 3G network and characterize mobility-induced channel state trends. We show that a user’s channel state is highly reproducible and leverage that to develop a data rate prediction mechanism. Our trace-based simulations of the (PF)2S Framework indicate that the framework can increase the throughput by 15%–55% compared to traditional PF schedulers, while improving fairness. Finally, we study fragmentation within a probability model of combinatorial structures. Our model does not refer to any particular application. Yet, it is applicable to dynamic spectrum access networks which can be used as the wireless access technology for numerous IoT applications. In dynamic spectrum access networks, users share the wireless resource and compete to transmit and receive data, and accordingly have specific bandwidth and residence-time requirements. We prove that the spectrum tends towards states of complete fragmentation. That is, for every request for j > 1 sub-channels, nearly all size-j requests are allocated j mutually disjoint sub-channels. In a suite of four theorems, we show how this result specializes for certain classes of request-size distributions. We also show that the delays in reaching the inefficient states of complete fragmentation can be surprisingly long. The results of this chapter provide insights into the fragmentation process and, in turn, into those circumstances where defragmentation is worth the cost it incurs.

Despite advances in monitoring spatiotemporal expression patterns of genes and proteins with fluorescent probes, direct detection of metabolites and small molecules remains challenging. Here we present an integrated circuit-based electrochemical camera chip capable of simultaneous spatial imaging of multiple redox-active metabolites, called phenazines, produced by Pseudomonas aeruginosa PA14 colony biofilms. Imaging of mutants with various capacities for phenazine production reveals local patterns of phenazine distribution in the biofilms. Such integrated circuit-based techniques promise wide applicability in detecting redox-active species from diverse biological samples.

Thesis (M.S.) Massachusetts Institute of Technology. Dept. of Electrical Engineering, 1946. / Bibliography: leaf 60. / by Sidney Alden Wingate. / M.S.