Physical and Control Layer Solutions for Reliable Machine-Type Random Access

Christopher Eric Boyd

Research output: ThesisDoctoral ThesisCollection of Articles

Abstract

Machine-type communications (MTC), in both the massive (mMTC) and ultra-reliable (uMTC) variety, will facilitate many new types of applications in 5G networks. Many of these applications will have service requirements that current wireless systems cannot meet. A particular challenge is providing reliable access to the large number of uncoordinated and infrequently active devices that characterise MTC. Restrictions on latencies preclude the traditional use of re-transmissions and lengthy resource scheduling procedures for reliability in the random access channel (RACH), encouraging instead a grant-free access paradigm. In such contention-based access, medium access control (MAC) layer coding schemes that provide repetition diversity are critical for limiting packet losses. This thesis considers Combinatorial Code Designs (CCD) for achieving the targets set for ultra-reliable low-latency communications (URLLC) in the 5G RACH. In contrast to coded access schemes which use random coding, we show that uniquely pre-allocating repetition patterns to users from a code designed specifically for use with successive interference cancellation (SIC) is a better solution for URLLC. We introduce interference cancelling (IC) codes that guarantee successful reception of all simultaneously accessing users up to a given number, which are shown to be particularly robust to packet loss when user activity is low, as in MTC. Simulations demonstrate that these codes outperform ALOHA-type schemes and meet the "five nines" reliability target of URLLC in both a simple collision model and a more realistic scenario that models both the MAC and physical (PHY) layers. Fully characterising the impact of the PHY layer uplink waveform on reliability is also paramount in designing an ultra-reliable random access procedure for 5G. As such, this thesis considers measures of time-frequency localisation (TFL) for stochastic signals. Such measures are complementary to those traditionally used in prototype filter design. We derive a generalised Heisenberg measure for multi-dimensional stochastic signals, consider the specific case of Gabor systems, and discern relevant bounds. The generalised TFL measure describes how well multiplexed waveform packets are contained inside their time-frequency resources, and thus their inter-user interference potential in a multiple-access scenario. We confirm the measure's veracity by simulating an asynchronous, grant-free random access system, and comparing the performance of small packets of common waveforms. At lower user activity, where out-of-band emissions are the limiting factor to reliability, the respective performances of these waveforms is consistent with the presented theory.
Translated title of the contributionPhysical and Control Layer Solutions for Reliable Machine-Type Random Access
Original languageEnglish
QualificationDoctor's degree
Awarding Institution
  • Aalto University
Supervisors/Advisors
  • Tirkkonen, Olav, Supervising Professor
  • Vehkalahti, Roope, Thesis Advisor
Publisher
Print ISBNs978-952-60-8417-6
Electronic ISBNs978-952-60-8418-3
Publication statusPublished - 2019
MoE publication typeG5 Doctoral dissertation (article)

Keywords

  • ultra-reliable low-latency communications
  • random access
  • combinatorial code designs
  • interference cancellation
  • time-frequency localisation

Fingerprint Dive into the research topics of 'Physical and Control Layer Solutions for Reliable Machine-Type Random Access'. Together they form a unique fingerprint.

Cite this