PhD Seminar: Performance Limits of Microwave and Dual Microwave/Millimeter Wave Band Networks

Monday, November 25, 2019 9:30 am - 9:30 am EST (GMT -05:00)

Candidate: Subhajit Majhi

Title: Performance Limits of Microwave and Dual Microwave/Millimeter Wave Band Networks

Date: November 25, 2019

Time: 9:30 AM

Place: E5 5047

Supervisor(s): Mitran, Patrick

Abstract:

In the last decade, fueled by an ever growing demand for wireless services, wireless communication technologies have advanced at an astonishing rate. In particular, in the 4th generation of cellular networks (4G), the user quality-of-service was vastly improved, and as a result, a plethora of new bandwidth-hungry applications such as high definition video streaming, virtual reality, etc., have become popular. While wireless standards are constantly evolving to meet the ever increasing demand for bandwidth, due to the vast number of users and services the available microwave (sub-6 GHz) spectrum has become scarce and is expected to be unable to support the rising service demands. To tackle this problem among others, the 5th generation of cellular networks (5G) was proposed, where incorporating additional spectrum from the millimeter-wave (mm-wave) band for wireless communication is emerging as a promising solution to the problem of microwave spectrum-scarcity.

Transmissions in the mm-wave band are typically achieved with highly directional steerable antenna arrays to counter the ill-effects of severe path-loss in such high operating frequencies. The resulting transmissions can thus be made to produce negligible interference at unintended receivers. Therefore, a key modeling feature of mm-wave transmissions are that they can often be modeled as point-to-point links. However, transmissions in the mm-wave band are inherently unreliable compared to those in the microwave band. Therefore, communicating simultaneously over both bands in an integrated mm-wave/microwave dual-band setup is emerging as a viable technology to enable significant improvements in quality-of-service in 5G while offsetting the shortcomings of either band. In this dual-band setting, high-rate data traffic can be carried by relatively unreliable high-bandwidth mm-wave links, while control signals and moderate-bandwidth traffic can be communicated over the reliable microwave band.

In the dual-band setting, while mm-wave transmissions are point-to-point in nature, in the underlying microwave band, transmissions intended for a specific receiver is typically received by other neighboring receivers as well. In conventional microwave networks, this leads to inter-user interference which presents a major performance bottleneck if not dealt with appropriately.   On the other hand, relay-aided cooperation is an important technology in conventional microwave networks, which has been shown to offset the adverse effects of multi-path fading. In this case, a relay node helps by boosting the transmission intended for a receiver, thereby extending the range and quality of the transmission.

In this thesis, we study two dual-band multi-user networks that deal with two important aspects of wireless communication: inter-user interference and relay-cooperation. The broad goal of this study is to characterize information-theoretical performance limits of such networks, which can then be used to obtain insights on the optimal encoding/decoding strategy, effective resource allocation schemes, etc.

In the first part of this thesis, we study a two-transmitter two-receiver dual-band Gaussian interference channel (IC) operating over an integrated mm-wave/microwave dual-band. This channel models a setting where a pair of single-transmitter single-receiver links communicate simultaneously, and thus mutually interfere. Here, transmissions in the underlying microwave band are modeled as a two-user conventional Gaussian IC (GIC). In contrast, a transmitter in the mm-wave band is assumed to be capable of communicating to either the desired destination or the interfered destination via a direct-link or a cross-link, respectively, created by beamforming to the appropriate receiver via a co-phased antenna-array. As such, a mm-wave transmitter can choose to convey fresh information to its desired destination via the direct-link, or to forward interference information to the interfered destination via the cross-link.

The dual-band IC is first classified into 3 classes according to whether the underlying microwave GIC has strong, weak, or mixed interference. Then, sufficient conditions over channel gains of both bands are obtained under which the capacity region of the 3 classes of the dual-band IC is characterized, and thus the optimal interference mitigation strategy for each case is identified. For cases in which the sufficient conditions do not necessarily hold in the dual-band IC, approximate capacity results are obtained that characterizes the capacity region to within 1/2 bit per channel use per user.  

The performance of the dual-band IC is likely to be impacted significantly by the point-to-point nature and large bandwidth of the mm-wave links, and specifically by whether the mm-wave spectrum is used as direct-links or cross-links. Transmitting in either the direct-links only or the cross-links only is not optimal for all channel conditions, and thus there exists a non-trivial trade-off between the two modes. Hence, to characterize the optimal performance of the dual-band IC over the mm-wave band parameters, we study the power allocation scheme over the mm-wave direct and cross-links that maximizes the sum-rate performance of the dual-band IC. The resulting optimal power allocation strategy is characterized in closed form, which possesses rich properties and reveals useful insights into the trade-offs in such networks.

In the second part of this thesis, we study a fading Gaussian multiple-access relay channel (MARC) over an integrated mm-wave/microwave dual-band, where two sources communicate to a destination with the help of a relay. In the dual-band MARC, transmission in the underlying microwave band is modeled as a conventional Gaussian MARC. However, similar to that in the dual-band IC,  a mm-wave transmitter in the dual-band MARC is modeled as being able to communicate to either the destination or the relay by creating a direct-link or a relay-link, respectively, by beamforming towards the appropriate receiver. For dual-band MARC, we characterize an achievable region and a set of rate upper bounds, and then obtain sufficient conditions over the channel gains of both bands under which its capacity region  is characterized in closed form.

Similar to the dual-band IC, the performance of the dual-band MARC will likely be significantly affected by whether the mm-wave spectrum is used as direct-links or relay-links. In fact, transmitting only in either of the two types of the mm-wave links is not optimal for all channel conditions, and hence a non-trivial trade-off between the two modes exists. To understand the impact of this trade-off, we study the transmission power allocation scheme over the mm-wave direct and relay-links that maximizes the sum-rate of the dual-band MARC. The resulting power allocation scheme, characterized in closed form, is observed to have rich structural properties, which reveal interesting insights into the trade-offs in relay cooperation in dual-band networks.

While integrated dual-band communication is emerging as a promising technology for 5G, currently a bulk of the connectivity is still supported by conventional microwave networks. However, the problem of interference mitigation for conventional interference networks is still not completely solved even for the basic case of a two-user IC. Motivated by this, in the third part of the thesis, we study the performance limits of the multiple-access interference channel (MAIC) that operates solely over the conventional single band. In the MAIC, a pair of two-user multiple access channels communicate over a shared medium and thus mutually interfere. Focusing on the weak interference case, which provides a more practical model of the inter-cell interference, we characterize an achievable strategy for the MAIC and 3 novel upper bounds on the sum-rate in the partially symmetric case, thereby providing improved sum-rate upper and lower bounds in these cases.