Ever-increasing data generation and transmission demands have been driving great advancements in wireline communications from both electrical and optical approaches for short- and long-distance scenarios, respectively. In the meter range scenario, both electrical and optical approaches face great challenges. That is, the lossy and bandwidth limited channels are the bottlenecks of electrical interconnect. For optical interconnect, the features, such as the required complex fabrication and the high environment sensitivity, increase the power and cost budgets significantly, making it uneconomical for short-distance communications. To mitigate these issues, low-loss dielectric channel-based interconnects have been investigated and demonstrated. However, all the design reported are all for point-to-point configurations, not suitable for multi-drop distributed architectures.

Besides CMOS transmitters and receivers, advanced spatial multiplexing schemes are fully investigated and architectures of wireline communication system at sub-THz/THz are selected according to the application scenarios. This dissertation also investigates the most used passive and active, on-chip and off-chip components, circuit modeling, layout optimization, and design strategies, such as neutralization and power maximization methods.

Two sub-THz interconnect systems, FDM based dual-band sub-THz interconnect and MDM multi-drop sub-THz interconnect, are proposed and demonstrated.

A low-loss and wideband Si DWG coupled with a pair of diplexers is employed to support two highly isolated and low loss sub-channels simultaneously. The proposed sub-THz interconnect achieves the energy efficiency of 1.58 pJ/b with the aggregate data rate of 22.6 Gb/s and BER better than 1e-12. It demonstrates the record bandwidth density of 150.7 Gb/s/mm2. Channelization provides a venue to boost the interconnect key metric of bandwidth density by taking full advantages of the abundant THz spectrum resource.

The theory of DWG mode coupler is derived and explained, and it can guide the multi-mode multi-drop waveguide design and provide optimization strategies effectively. Then a multi-drop sub-THz interconnect system is demonstrated, enabling three simultaneous logical channels for E11^y, E21^y and E31^y mode and supporting data rates of 24 Gb/s, 22 Gb/s and 19 Gb/s, respectively, with the BER better than 1e-12. The demonstrated aggregate data rate of the three channels is 65 Gb/s with the energy efficiency of 1.6 pJ/b.

To the authors’ knowledge, This is the first time to demonstrate multi-mode multi-drop DWG based interconnect. One note to make is that although the demonstration channel length is 5.2 cm, this is due to the size constraint of the wafer used to fabricate the channel. This interconnect system can be readily extended to the meter range due to the channel ultra-low loss feature. Furthermore, it can also scale to more modes to support more logic channels per physical link and can be extended to multi-dimension, two-/three- dimension, interconnect systems. Besides, with more advanced semiconductor technologies for active circuits, the data rate per channel will be further increased. Therefore, we believe that demonstrated multi-mode multi-drop sub-THz interconnect systems open a new path with high potentials to complement the existing electrical and optical interconnect to address the challenging meter range wireline communication scenarios.

In my dissertation, I explore questions about strategic decision making and interactions between market participants, with an emphasis on information transmission and information design. I apply game-theoretic models to study how strategic information transmission influences economic outcomes in trading environments.

In Chapter 1, I study the information provision role and the incentive provision role of certification in a moral hazard setting in a seller-induced certification model versus a buyer-induced certification model. My results explain empirical observations in the credit rating market and provide policy implications regarding the "issuer-pays" rating mode versus the "investor-pays" rating mode in the financial market.

Chapter 2 examines the incentive of a long-run seller to disclose previous offers in a dynamic market for lemons and identifies the impact of allowing voluntary disclosure on the market information structure. Compared with the models of mandatory disclosure and mandatory non-disclosure, the optional disclosure model generates a novel set of equilibria by allowing flexibility in the disclosure option. My result also proves the efficiency of optional disclosure, which explains a change in eBay's disclosure scheme about previous transactions.

In Chapter 3, I consider a long-term contracting problem between a monopolistic seller and a present-biased buyer with asymmetric information in a Markov environment. The buyer and the seller are fully aware of the degree of inconsistent discounting. I characterize the optimal contract and identify the novel impact of time inconsistency on the optimal allocations.