UC Santa Barbara
Millimeter wave picocellular networks: capacity analysis and system design
- Author(s): Marzi, Zhinus
- Advisor(s): Madhow, Upamanyu
- et al.
The explosive growth in demand for wireless mobile data, driven by the proliferation
of ever more sophisticated handhelds creating and consuming rich multimedia, calls for
orders of magnitude increase in the capacity of cellular data networks. Millimeter wave
communication from picocellular base stations to mobile devices is a particularly promising
approach for meeting this challenge because of two reasons. First, there is a large
amount of available spectrum, enabling channel bandwidths of the order of Gigahertz
(GHz) which are 1-2 orders of magnitude higher than those in existing WiFi and cellular
systems at lower carrier frequencies. Second, the small carrier wavelength enables the
realization of highly directive steerable arrays with a large number of antenna elements,
in compact form factors, thus significantly enhancing spatial reuse. Hence, we propose to
employ the 60 GHz unlicensed band for basestation to mobile communication in outdoor
We first investigate the basic feasibility of such networks, showing that 60GHz links
are indeed viable for outdoor applications. For this purpose, we provided link budget
calculations along with preliminary simulations which show that despite the common
concerns about higher oxygen absorption and sensitivity to movement and blockage,
picocloud architecture provides availability rate of more than 99%.
Next, we explore the idea of increasing spatial reuse by shrinking picocells hoping
that interference is no longer the bottleneck given the highly directive antenna arrays at
this band. Our goal is to estimate the achievable capacity for small picocells along an urban canyon. We consider basestations with multiple faces or sectors, each with one or more antenna arrays. Each such array, termed subarray can employ Radio Frequency
(RF) beamforming to communicate with one mobile user at a time. We first focus on
characterization and modeling the inter-cell interference for one subarray on each face.
Our analysis provides a strong indication of very large capacity (in the order of Tbps/km)
with a few GHz of bandwidth.
Following this, we explore the impact of adding multiple subarrays per face. This leads
to intra-cell interference as well as additional inter-cell interference. While the effect of
additional inter-cell interference can be quantified within our previous framework, intracell
interference has inherently different features that call for new approaches for analysis
and design. We propose a cross-layer approach to suppress the intra-cell interference in
two stages: (a) Physical layer (PHY-layer) method which mitigates interference by joint
precoding and power adaptation and (b) Medium Access Control layer (MAC-layer)
method which manages the residual interference by optimizing resource allocation. We
then estimate the capacity gain over conventional LTE cellular networks and establish
that 1000-fold capacity increase is indeed feasible via mm-wave picocellular networks.
Lastly, we examine fundamental signal processing challenges associated with channel
estimation and tracking for large arrays, placed within the context of system design
for a mm-wave picocellular network. Maintainance of highly directive links in the face
of blockage and mobility requires accurate estimation of the spatial channels between
basestation and mobile users. Here we develop the analytical framework for compressive
channel estimation and tracking. We also address the system level design discussing
link budget, overhead, and inter-cell beacon interference. Simulation results demonstrate
that our compressive scheme is able to resolve mm-wave spatial channels with a relatively
small number of compressive measurements.