PaperlayerView exact revision on arXiv

Source-linked AI summary

Comparative Analysis of Initial Access Techniques in 5G mmWave Cellular Networks

Marco Giordani, Marco Mezzavilla, Nicolas Barati, Sundeep Rangan, Michele Zorzi

arXiv:1605.00101v1cs.NI

TL;DR

5G mmWave systems require directional transmission to overcome propagation loss, making initial access dependent on finding suitable BS–UE directions and potentially increasing access delay. The paper surveys and compares exhaustive and iterative techniques using misdetection probability and discovery delay, finding that the preferred strategy depends on target SNR, channel setting, and user-density conditions.

  • Problem

    High mmWave pathloss requires directional BS and UE transmissions, but finding initial directions complicates and can delay cellular initial access.

  • Method

    The paper surveys recent directional initial-access techniques and compares exhaustive and iterative search using misdetection probability and discovery delay under modeled channel and overhead conditions.

  • Results

    The preferred initial-access strategy depends on target SNR regime, while exhaustive procedures are preferred for minimum edge-user coverage in dense urban multipath channels.

  • Takeaways & Limitations

    Iterative techniques reduce angular-search slots but have higher misdetection probability, whereas exhaustive techniques are preferred for dense networks requiring edge-user coverage.

  • Takeaways & Limitations

    Iterative procedures are not recommended for very dense networks with multiple users in different macro sectors, because sequential refinement can lengthen discovery.

Abstract

from arXiv · show

The millimeter wave frequencies (roughly above 10 GHz) offer the availability of massive bandwidth to greatly increase the capacity of fifth generation (5G) cellular wireless systems. However, to overcome the high isotropic pathloss at these frequencies, highly directional transmissions will be required at both the base station (BS) and the mobile user equipment (UE) to establish sufficient link budget in wide area networks. This reliance on directionality has important implications for control layer procedures. Initial access in particular can be significantly delayed due to the need for the BS and the UE to find the initial directions of transmission. This paper provides a survey of several recently proposed techniques. Detection probability and delay analysis is performed to compare various techniques including exhaustive and iterative search. We show that the optimal strategy depends on the target SNR regime.

I. INTRODUCTION

5G mmWave systems use abundant spectrum to support demanding throughput, latency, reliability, connectivity, and energy goals, but directional transmission makes initial access challenging. The paper surveys and compares directional search schemes using misdetection probability and discovery delay under varying channel conditions and overhead constraints.

  • mmWave bands above 10 GHz offer largely unused spectrum that can support higher data rates in future mobile broadband networks.
  • Directional transmission complicates initial access because BSs and UEs must determine suitable initial transmission directions before establishing a link.Unlike current LTE, where initial access can use omni-directional channels, mmWave access requires directional search.
  • The paper surveys recent directional initial-access techniques and compares exhaustive search with an iterative scheme that narrows the search direction successively.
  • The comparison evaluates misdetection probability and discovery delay under overhead constraints and varying channel conditions.
  • The paper reports that the preferred strategy depends on the target SNR regime.

II. RELATED WORK

Prior mmWave initial-access research includes exhaustive scanning, directional cell discovery, hierarchical and adaptive beam training, random beamforming, and low-complexity beam selection. This paper positions its contribution as a comparison of multiple procedures for coverage probability and delay in realistic dense urban multipath scenarios.

  • Prior work studies sequential 360° scanning, periodic synchronization signals in directional or random directions, and alternative scanning and signaling procedures.
  • Other proposals examine random directional beamforming, low-complexity beamforming, hierarchical procedures, energy-efficient link configuration, and adaptive beam training.
  • Beam selection has also been addressed through analog beamformers and compressive sensing without explicit channel estimation.
  • This paper compares multiple initial-access procedures by coverage probability and delay in a realistic dense, urban, multipath scenario.

III. INITIAL ACCESS IN 5G-MMWAVE NETWORKS

The study adopts a periodic synchronization-signal slot structure and evaluates two modified initial-access schemes inspired by prior exhaustive and hierarchical approaches.

  • The simulations use a slot structure in which the PSS is transmitted periodically for a signal duration within each transmission.The work considers potentially different signal periods and defines simulation parameters in Section IV.

A. Exhaustive search

Exhaustive search sequentially scans the full angular space using predefined BS directions and UE combining directions, then selects the beam pair with the highest received SNR. The procedure trades search duration for complete angular coverage and final beam selection after all directions are scanned.

  • Exhaustive search performs brute-force sequential beam searching over a codebook of N directions covering the whole angular space.
  • The BS transmits narrow-beam messages across different slots while the UE configures directional reception and reports signals whose SNR exceeds a fixed threshold.
  • The implementation uses a 64-antenna BS with N = 16 directions, while the UE can receive through either 4 wide beams or 8 narrower beams.
  • For four-direction reception, each BS direction uses 4 consecutive DL slots followed by 1 UL feedback slot, forming one initial-access macroelement.
  • The BS and UE repeat directional scanning until the whole angular space is covered.
  • After all 16 BS directions are scanned, both sides select the beam associated with the highest received SNR.

B. Iterative search

Iterative search divides initial access into a coarse scan across four macro directions and a refinement scan within the selected sector. The BS and UE retain their best directions, then refine the selected macro sector using narrow beams.

  • B. Iterative search: Iterative search scans the angular space in two stages: four macro wide beams first, followed by refinement within the selected sector.The BS and UE identify their best directions after the first phase, and the BS then limits the second phase to that macro sector.
  • B. Iterative search: The second phase repeats the refining procedure across N2 = 4 narrow sectors within the selected macro sector.The UE uses its previously identified best receiving direction while the BS inspects the refining directions.
  • B. Iterative search: The first phase repeats the macroelement procedure across N1 = 4 wide sectors covering the full angular space.Each macroelement includes downlink synchronization transmissions and an uplink response before the next direction is scanned.
  • B. Iterative search: After refinement, the BS selects the narrow beam with the highest saved SNR, leaving both endpoints with their best directional configuration.The procedure ends when the BS and UE know how to directionally reach each other.

IV. SIMULATION MODEL AND NUMERICAL RESULTS

The simulations model static, dense urban mmWave deployments using measurement-based 28 GHz channel statistics and analog planar arrays. Performance is evaluated through discovery delay and misdetection probability for different UE reception configurations.

  • IV. SIMULATION MODEL AND NUMERICAL RESULTS: The simulations assume a static deployment with no user motion, handover management, or UE motion tracking.The parameters are intended to represent realistic system design considerations.
  • IV. SIMULATION MODEL AND NUMERICAL RESULTS: The channel model uses 28 GHz New York City measurements with distance-based LOS, NLOS, and outage pathloss conditions.It also models spatial clusters and small-scale fading for dense urban mmWave micro- and picocellular networks.
  • IV. SIMULATION MODEL AND NUMERICAL RESULTS: The modeled channel suggests that good performance for most users may require cell radii smaller than the envisioned 100 ÷ 200 m range.This conclusion is stated specifically for the channel described in the cited measurement-based model.
  • IV. SIMULATION MODEL AND NUMERICAL RESULTS: The study uses analog beamforming with Uniform Planar Arrays at both the BS and UE, using 8 × 8, 4 × 4, or 2 × 2 elements.The antenna-element spacing is λ/2, and the array patterns were selected for small-cell urban deployment.
  • IV. SIMULATION MODEL AND NUMERICAL RESULTS: The SNR threshold is τ = −5 dB; signals below this threshold are treated as undetected by the UE.The signal duration is Tsig = 10 µs, while beam switching is considered negligible relative to that duration.
  • IV. SIMULATION MODEL AND NUMERICAL RESULTS: The evaluation measures discovery delay and misdetection probability for UEs receiving through either 4 or 8 beams.Misdetection probability records the fraction of experiments in which the perceived SNR falls below threshold.

A. Required number of slots

Discovery delay is determined by the number of slots required by each initial-access procedure under the specified signal duration and overhead. Iterative techniques reduce delay by refining only the selected macro sector instead of scanning the entire angular space.

  • A. Required number of slots: With Tsig = 10 µs and φov = 5%, the minimum interval between slot transmissions is Tper = 200 µs.The interval follows from Tper = Tsig/φov.
  • A. Required number of slots: Discovery delay is computed as Ns · Tsig/φov, where Ns is the number of slots required by the initial-access procedure.This formulation accounts for the required slot transmissions and the target overhead.
  • A. Required number of slots: Iterative techniques outperform exhaustive techniques in discovery delay because they refine only one macro sector rather than scanning the full 360° angular space.The comparison is reported for the discovery-delay metric in Table V.

B. Misdetection probability

Misdetection depends on beamforming gain, distance, and signal duration. Iterative search generally has higher misdetection than exhaustive search, while outage conditions can sharply worsen both.

  • Iterative search has higher misdetection probability than exhaustive search at the same distance because its first phase uses only four antenna elements.Fewer antennas reduce beamforming gain and perceived SNR, increasing the chance that SNR falls below the threshold.
  • Figure 2 compares PMD for exhaustive and iterative searches as signal duration Tsig varies at a 95-meter BS–UE distance.
  • At 0–30 meters, nearly all algorithms achieve acceptable misdetection probability because line-of-sight conditions are likely and pathloss remains small.
  • At 100–200 meters, nearly all algorithms show unacceptable misdetection probability values.
  • Around 150 meters, outage pathloss sharply reduces perceived SNR even with exhaustive beamforming, causing a sudden increase in PMD.

C. Trade-off between delay and PMD (total delay)

Reducing misdetection for edge users requires longer signals for iterative search, while exhaustive search can use shorter slots. In dense networks, iterative refinement can further increase discovery delay and worsen misdetection.

  • Total delay accounts for both the slots required by each IA technique and the PMD specification, with Tper selected to maintain 5% overhead.
  • Table VI reports total delay for guaranteeing PMD < 0.01 for edge users.
  • Iterative techniques require very long signals to overcome low beamforming gain, whereas exhaustive searches can use shorter slots.
  • For cells around 100 meters, iterative search has higher total delay than exhaustive search when good edge-user coverage is required.
  • In very dense networks, iterative refinement can take longer than exhaustive search and still have worse misdetection probability when multiple users occupy different macro sectors.

V. CONCLUSIONS AND FUTURE WORKS

The analysis identifies a delay–misdetection trade-off between iterative and exhaustive initial-access techniques. For dense urban multipath channels and edge-user coverage near 100 meters, exhaustive procedures are preferred because they achieve smaller total delay.

  • The study compares initial-access implementations for 5G millimeter-wave cellular networks, where directionality is needed during initial synchronization and access.
  • Iterative techniques require fewer search slots but have higher misdetection probability because their first phase uses a small antenna array.
  • For dense urban multipath channels and edge users around 100 meters from the BS, exhaustive procedures are preferred because they achieve smaller total delay.
  • Future work will investigate digital or hybrid beamforming to steer multiple narrow beams simultaneously, alongside HetNet- and context-information-assisted initial access.
Loading 1605.00101v1…