1.1 What is RF Planning?

Radio Frequency (RF) planning is the engineering discipline of designing wireless cellular networks to deliver optimal coverage, capacity, and quality of service to subscribers. It is the foundational step that determines where base stations are placed, how they are configured, and how the radio resources are managed across the network.

An RF planner must balance three competing objectives that form the "RF Planning Triangle": maximizing coverage area, providing sufficient capacity for traffic demand, and maintaining acceptable quality (measured by signal-to-interference-plus-noise ratio, SINR). These three goals are often in tension — wider coverage typically means fewer sites but less capacity, while dense deployments improve capacity but increase interference.

1.2 Evolution of Cellular Networks

To understand modern RF planning, we must appreciate how cellular technology has evolved over four decades. Each generation brought fundamental changes to how radio networks are designed:

• 1G (1980s): Analog FM systems (AMPS, NMT, TACS). Cell sizes of 2–20 km. Planning focused purely on coverage using simple Okumura-Hata models. No frequency reuse optimization.
• 2G (1990s): Digital systems (GSM, IS-95/CDMA). Introduced frequency reuse patterns (4/12, 3/9, 1/3), power control, and the concept of capacity-limited vs. coverage-limited planning.
• 3G (2000s): WCDMA/CDMA2000. Wideband spreading with universal frequency reuse (reuse-1). Soft handover zones, pilot pollution management. Planning shifted to interference management.
• 4G LTE (2010s): OFDMA downlink, SC-FDMA uplink. Flat architecture, all-IP. Inter-Cell Interference Coordination (ICIC), carrier aggregation. Planning became multi-band, multi-layer.
• 5G NR (2020s): Flexible numerology, massive MIMO, mmWave. Beam-based coverage, TDD slot configuration. Planning requires 3D beam management and new propagation models for FR2.

1.3 The RF Planning Workflow

RF planning follows a structured workflow that moves from high-level requirements to detailed site-specific configurations. The process is iterative — results from each phase feed back into previous phases for refinement.

Phase 1: Requirements & Dimensioning

The process begins with understanding the deployment area, subscriber forecasts, traffic models, and coverage/capacity targets. Using these inputs, the RF planner performs network dimensioning to estimate the total number of sites required. This involves link budget calculations to determine the maximum cell radius and capacity analysis to determine the minimum site density.

Phase 2: Nominal Planning

In nominal planning, candidate site locations are identified based on terrain, clutter, population density, and coverage requirements. Using propagation models calibrated to the local environment, coverage predictions are generated for each candidate. The output is a nominal site plan showing approximate tower locations, antenna heights, and expected coverage footprints.

Phase 3: Detailed Planning

Each nominal site undergoes detailed planning, where exact antenna configurations (type, height, azimuth, tilt), frequency assignments, power settings, and neighbor relations are defined. Tools like Atoll, ASSET, or Planet are used for Monte Carlo simulation of coverage and capacity. Physical Cell Identity (PCI), PRACH root sequences, and tracking area codes are allocated.

Phase 4: Optimization

After the network is deployed, drive testing is performed to validate coverage predictions. Propagation models are tuned using measured data. Parameters are optimized iteratively based on KPIs — coverage holes are filled, interference is mitigated, and handover parameters are fine-tuned.

1.4 Key 3GPP & ITU Standards for RF Planning

RF planning is governed by a comprehensive set of international standards. The two primary bodies are 3GPP (3rd Generation Partnership Project), which defines the radio access technology specifications, and ITU-R (International Telecommunication Union — Radiocommunication Sector), which defines propagation models and spectrum regulations.

1.5 The Role of the RF Engineer

The RF planning engineer sits at the intersection of physics, engineering, economics, and operations. Their responsibilities span the entire network lifecycle:

• Pre-Launch: Network dimensioning, site selection, coverage design, frequency planning, parameter definition
• Launch: Drive test validation, model tuning, first-pass optimization, KPI benchmarking
• Steady-State: Capacity monitoring, interference management, handover optimization, new site integration
• Evolution: Technology upgrades (4G→5G), new band deployment, massive MIMO activation, DSS planning

1.6 Book Roadmap

This book is organized into five parts that mirror the natural progression of RF planning knowledge:

• Part I (Chapters 1–5): Foundations — propagation physics, models, antennas, and spectrum
• Part II (Chapters 6–10): 4G LTE RF planning — link budget, capacity, coverage, interference, physical layer
• Part III (Chapters 11–16): 5G NR RF planning — link budget, capacity, massive MIMO, beamforming, mmWave
• Part IV (Chapters 17–24): Advanced topics — indoor, dimensioning, site engineering, drive testing, optimization, HetNets, special scenarios
• Part V (Chapters 25–27): Tools, AI/ML, and future technologies (5G-Advanced, 6G)

2.1 Electromagnetic Wave Propagation

Radio waves are electromagnetic (EM) waves that propagate through space at the speed of light (c = 3 × 10⁸ m/s). The relationship between frequency (f), wavelength (λ), and the speed of light is given by:

At cellular frequencies, the wavelength ranges from approximately 15 cm at 2 GHz (4G LTE) to 5 mm at 60 GHz (5G mmWave). This wavelength determines how radio waves interact with the environment — shorter wavelengths suffer higher attenuation but can exploit smaller antenna elements for beamforming.

2.2 Free-Space Path Loss (FSPL)

In an ideal environment with no obstacles, the signal power decreases with the square of the distance from the transmitter. This is the Free-Space Path Loss (FSPL), defined by ITU-R P.525:

2.3 Propagation Mechanisms

In the real world, radio signals encounter obstacles that modify the signal through four primary mechanisms:

2.3.1 Reflection

When a radio wave strikes a surface that is large compared to the wavelength (buildings, ground, water), the wave is reflected according to Snell's law. The reflected signal can constructively or destructively interfere with the direct signal. At cellular frequencies, buildings and the ground surface are the primary reflectors.

2.3.2 Diffraction

When a radio wave encounters an obstacle edge (rooftop, hilltop, building corner), it bends around the obstacle, allowing signal to reach into the shadow region behind it. This is the primary mechanism enabling coverage in non-line-of-sight (NLOS) conditions. Diffraction loss increases with frequency, which is why low-band signals "bend" better around obstacles than mmWave.

2.3.3 Scattering

When a wave hits objects that are comparable to or smaller than the wavelength (foliage, street signs, lamp posts, rough surfaces), the energy is scattered in multiple directions. Scattering is particularly significant at higher frequencies and is a major propagation mechanism for mmWave signals in urban environments.

2.3.4 Absorption

Materials absorb RF energy and convert it to heat. The amount of absorption depends on the material and frequency. Concrete walls typically attenuate signals by 10–25 dB, while low-E glass can add 20–40 dB of loss. At mmWave frequencies, atmospheric gases (oxygen at 60 GHz, water vapor at 22 GHz) cause additional absorption per ITU-R P.676.

2.4 Multipath Propagation & Fading

In a real radio environment, the received signal is the sum of multiple copies of the transmitted signal arriving via different paths (direct, reflected, diffracted, scattered). These copies arrive with different delays, amplitudes, and phases. When they combine at the receiver, the result is multipath fading — rapid fluctuations in signal strength that can vary by 30–40 dB over distances as short as half a wavelength.

2.4.1 Small-Scale Fading

Small-scale fading describes rapid signal variations over short distances (on the order of wavelength). Two statistical models characterize this:

• Rayleigh Fading: When there is no dominant line-of-sight (LoS) path, the signal envelope follows a Rayleigh distribution. This is typical for urban NLOS conditions. Deep fades of 20–30 dB below the mean are common.
• Rician Fading: When a strong LoS component exists alongside scattered components, the envelope follows a Rician distribution characterized by the K-factor (ratio of LoS power to scattered power). Higher K means less fading depth.

2.4.2 Large-Scale Fading (Shadow Fading)

Large-scale fading, or shadow fading, describes slow signal variations over distances of tens to hundreds of meters, caused by large obstacles (buildings, hills) blocking the signal path. Shadow fading follows a log-normal distribution and is characterized by its standard deviation (σ), typically 6–10 dB in urban environments. In link budgets, a shadow fading margin is added to ensure coverage at the cell edge with a specified probability (e.g., 8.2 dB for 90% edge reliability with σ = 8 dB).

2.5 Doppler Effect

When a mobile user is moving, the received frequency shifts due to the Doppler effect. The maximum Doppler shift is:

The Doppler spread affects the coherence time of the channel, which determines how quickly the channel changes. This is critical for choosing the right subcarrier spacing in 5G NR — higher speeds require wider subcarrier spacing to combat inter-carrier interference (ICI).

2.6 Penetration Loss

One of the most critical parameters in RF planning is building penetration loss (BPL). This determines how much signal is lost when it enters a building from outside. ITU-R P.2109 provides a statistical model for building entry loss that varies with frequency.

2.7 Fresnel Zone & Line of Sight

For a radio link to achieve near free-space propagation, not only must there be a direct line of sight (LoS) between transmitter and receiver, but the first Fresnel zone must be substantially clear of obstructions. The first Fresnel zone is an ellipsoid around the direct path whose radius at any point is:

2.8 Chapter Summary

3.1 Classification of Propagation Models

Propagation models predict the path loss between a transmitter and receiver based on distance, frequency, antenna heights, and environmental characteristics. They fall into three broad categories:

• Empirical Models: Based on extensive measurement campaigns. Fast to compute but limited to the environments where measurements were taken. Examples: Okumura-Hata, COST-231 Hata, SUI.
• Semi-Empirical (Semi-Deterministic): Combine measured data with simplified physics (diffraction over rooftops, street canyons). Better accuracy in specific morphologies. Examples: COST-231 Walfish-Ikegami, Ericsson 9999.
• Deterministic Models: Use detailed 3D building/terrain data and ray-tracing or FDTD to compute path loss from first principles. Highest accuracy but computationally expensive. Used for indoor planning and mmWave.

3.2 Okumura-Hata Model

The most widely used empirical model in cellular planning, based on Okumura's extensive measurements in Tokyo (1968) and Hata's mathematical formulation (1980). Valid for:

• Frequency: 150–1500 MHz (extended to 2000 MHz by COST-231)
• Distance: 1–20 km
• BS height: 30–200 m
• UE height: 1–10 m

3.3 COST-231 Hata Model

The European COST-231 committee extended the Hata model to cover the 1500–2000 MHz range, making it suitable for DCS-1800 and early UMTS planning:

3.4 COST-231 Walfish-Ikegami Model

This semi-empirical model separates the path loss into three components: free-space loss, rooftop-to-street diffraction, and multi-screen diffraction. It accounts for street width, building height, building separation, and street orientation angle. Valid for 800–2000 MHz, distances 0.02–5 km.

3.5 ITU-R Propagation Models

3.5.1 ITU-R P.1812 — Point-to-Area Terrestrial Path Loss

P.1812 is the ITU's recommended model for terrestrial point-to-area coverage prediction. It uses detailed terrain profile data and supports frequencies from 30 MHz to 6 GHz. The model combines diffraction (Delta-Bullington method), tropospheric scatter, ducting, and clutter losses into a comprehensive prediction. It is the basis for modern planning tools like CellScope Pro.

3.5.2 ITU-R P.1411 — Short-Range Outdoor Propagation

Designed for small cells and street-level propagation at 300 MHz to 100 GHz. Covers LoS, NLoS, and over-rooftop scenarios. Critical for 5G small cell planning in urban environments.

3.5.3 ITU-R P.2109 — Building Entry Loss

Provides a statistical model for building penetration loss as a function of frequency and building type (traditional vs. thermally efficient). The median building entry loss at 3.5 GHz for a thermally efficient building is approximately 22 dB — a critical parameter for indoor coverage planning from outdoor cells.

3.6 3GPP TR 38.901 — 5G Channel Model

The definitive propagation model for 5G NR planning, covering 0.5 to 100 GHz across all deployment scenarios. TR 38.901 defines path loss formulas, LoS probability, and large-scale parameters for multiple environments:

3.7 LoS Probability Models

TR 38.901 also defines the probability that a link is LoS vs. NLoS as a function of distance. This is critical for accurate Monte Carlo coverage simulations:

3.8 Model Calibration

No propagation model is accurate out-of-the-box for a specific deployment area. Model calibration (also called model tuning) adjusts the model coefficients using drive test measurements from the target area. The process:

• Step 1: Collect drive test data (RSRP measurements + GPS coordinates) from representative routes covering urban, suburban, and open morphologies.
• Step 2: Match each measurement point to the serving cell and compute the measured path loss: PL_meas = P_tx + G_ant - CableLoss - RSRP_meas
• Step 3: For each point, compute predicted path loss from the model: PL_pred
• Step 4: Minimize the error (PL_meas - PL_pred) using least-squares regression, adjusting the model's offset (K1) and slope (K2) parameters.
• Step 5: Validate that RMSE (root-mean-square error) between measured and predicted is below 8 dB for macro cells. Values below 6 dB indicate excellent calibration.

4.1 Antenna Types in Cellular Networks

Cellular networks use directional antennas to focus radio energy toward the intended coverage area. The primary antenna types are:

• Omni-directional: Radiates equally in all horizontal directions. Used for rural sites, indoor ceiling-mount, and IoT. Typical gain: 2–6 dBi.
• Sector (Panel): The workhorse of macro sites. Typically 65° horizontal beamwidth for 3-sector configuration. Gain: 15–18 dBi. Available in single-band, multi-band, and wideband variants.
• Active Antenna System (AAS): Combines radio unit and antenna into a single unit with multiple independently controlled elements. Enables massive MIMO beamforming. 32T32R or 64T64R configurations. Gain: 23–27 dBi (including beamforming gain).
• Small Cell Antenna: Compact antennas for urban street-level or indoor deployment. Omni or low-gain directional. Gain: 5–10 dBi.

4.2 Key Antenna Parameters

4.3 Antenna Tilt — The RF Planner's Primary Tool

Antenna tilt is the single most important parameter the RF planner controls after site selection. It determines the effective cell radius, cell-edge signal strength, and inter-cell interference levels.

• Mechanical Tilt (M-Tilt): Physically tilting the antenna bracket downward. Affects the entire pattern uniformly. Side lobes also tilt. Simple but requires tower climb to adjust.
• Electrical Tilt (E-Tilt / RET): Adjusting internal phase shifters to tilt the beam electronically. Only the main beam tilts; side lobes are suppressed. Can be adjusted remotely (RET = Remote Electrical Tilt via AISG interface).
• Combined Tilt: M-tilt + E-tilt. Best practice is to use E-tilt for fine adjustment (0–10°) and add M-tilt only when more than 10° total downtilt is needed.

4.4 Antenna Tilt Optimization — Science & Practice

4.4.1 Optimal Tilt Calculation

4.4.2 Tilt vs. Coverage vs. Interference Trade-off

4.4.3 Automated Tilt Optimization (SON)

Modern networks use Self-Organizing Network (SON) algorithms to continuously optimize antenna tilt based on live KPIs:

4.4.4 CCI Minimization Through Tilt

Co-Channel Interference (CCI) is the primary limiter of cell-edge performance. The relationship between tilt and CCI follows a predictable pattern:

• Uptilt/no-tilt: Antenna main beam overshoots the cell → strong signal reaches deep into neighbor cells → high CCI. Cell-edge SINR: -5 to +3 dB.
• Moderate downtilt (6–10°): Main beam contained within the cell → signal drops rapidly beyond cell edge → low CCI. Cell-edge SINR: +5 to +12 dB. Optimal for throughput.
• Excessive downtilt (>12°): Cell shrinks too much → coverage gaps appear between cells → low RSRP at cell edge even though SINR is good. Users at cell boundary have no signal.
• Rule of 6 dB: Every 1° of additional downtilt reduces the received signal at the neighbor cell by approximately 1–2 dB. A 3° tilt increase reduces CCI by ~4–6 dB.

4.5 Antenna Selection Guidelines

5.1 Spectrum Fundamentals

Radio spectrum is the most valuable and constrained resource in cellular networks. Operators acquire spectrum licenses through auctions, typically paying billions of dollars. The amount and type of spectrum directly determines the coverage and capacity a network can deliver.

5.2 4G LTE Frequency Bands

LTE operates in both FDD (Frequency Division Duplex) and TDD (Time Division Duplex) modes across numerous bands defined in 3GPP TS 36.104. Key global LTE bands:

5.3 5G NR Frequency Bands

5G NR defines two frequency ranges per 3GPP TS 38.104:

• FR1 (410 MHz – 7.125 GHz): Sub-6 GHz bands. Maximum channel bandwidth: 100 MHz. SCS options: 15, 30, 60 kHz. Used for wide-area coverage and capacity.
• FR2 (24.25 GHz – 52.6 GHz): Millimeter wave. Maximum channel bandwidth: 400 MHz. SCS options: 60, 120, 240 kHz. Used for extreme capacity in dense urban hotspots.

5.4 Channel Bandwidth & Numerology

5G NR introduces flexible numerology with variable subcarrier spacing (SCS). The SCS determines the symbol duration, which affects coverage (larger cells need longer CP) and Doppler resilience (high-speed UEs need wider SCS):

5.5 Carrier Aggregation & Dual Connectivity

To maximize throughput, operators combine multiple carriers:

• LTE CA: Up to 5 Component Carriers (CC), 100 MHz total. Intra-band (contiguous/non-contiguous) or inter-band.
• NR CA: Up to 16 CCs in FR1, aggregating up to 1.6 GHz of bandwidth.
• EN-DC (E-UTRA NR Dual Connectivity): LTE anchor + NR secondary cell group. Used in NSA (Non-Standalone) 5G deployments. LTE provides control plane; NR adds data plane capacity.
• NR-DC: NR + NR dual connectivity. Used in SA deployments to combine FR1 + FR2.

5.6 Dynamic Spectrum Sharing (DSS)

DSS allows LTE and NR to share the same carrier dynamically on a per-slot or per-subframe basis. This enables operators to launch 5G on existing LTE spectrum without a dedicated NR carrier. However, DSS has limitations:

• Reduced spectral efficiency (10–20% overhead for CRS/SSB coexistence)
• Limited NR bandwidth per slot (depends on LTE traffic load)
• Best used as a transition technology until dedicated NR spectrum is available

5.7 The Layered Spectrum Strategy

5.8 Chapter Summary

6.1 Link Budget Fundamentals

The link budget is the most fundamental calculation in RF planning. It determines the Maximum Allowable Path Loss (MAPL) between the base station and the UE. The cell radius is then derived by inverting the propagation model at this MAPL value.

6.2 Receiver Sensitivity

Receiver sensitivity is the minimum signal power at which the receiver can demodulate the signal with acceptable quality (BLER ≤ 10%). It depends on the thermal noise floor, receiver noise figure, and required SINR:

6.3 Complete LTE Link Budget — Worked Example

6.4 Complete Link Budget Table

7.1 LTE Throughput Calculation

The theoretical peak throughput of an LTE cell is determined by the bandwidth, MIMO configuration, modulation order, and coding rate:

7.2 Spectral Efficiency vs. SINR

7.3 Cell Capacity Estimation

7.4 Traffic Demand Estimation

Capacity dimensioning matches supply (cell throughput) with demand (subscriber traffic):

8.1 Coverage Objectives & Thresholds

Coverage planning starts with defining what "covered" means. In LTE, coverage is measured by RSRP (Reference Signal Received Power), RSRQ (Reference Signal Received Quality), and SINR:

8.2 Clutter Classification

The deployment area is classified into morphology types (clutter classes) that affect propagation loss. Each class has an associated clutter attenuation factor used in path loss calculation:

8.3 Site Selection Criteria

• Height: Antenna should be above average clutter height. Urban: 25–40 m, Suburban: 20–30 m, Rural: 30–50 m.
• Location: Near the center of the target coverage area. Avoid placing on the edge of a cliff or ridge (creates overshoot).
• Clear LoS: Unobstructed view toward target coverage area. No tall buildings or terrain features in the near field.
• Access: Road access for installation and maintenance. Power and backhaul availability.
• Co-location: Prefer existing tower sites to reduce cost and permitting time.
• EMF compliance: Meet regulatory RF exposure limits (typically 6-minute average exclusion zone).

8.4 Monte Carlo Simulation

Coverage planning tools like Atoll use Monte Carlo simulation to predict coverage statistics. The process: (1) place thousands of random "test mobiles" across the area, (2) calculate received signal from all surrounding cells for each point, (3) determine serving cell, SINR, and throughput, (4) compute coverage probability as the percentage of points meeting the threshold. Target: 95%+ area coverage for outdoor, 90%+ for indoor.

9.1 Inter-Cell Interference in LTE

LTE uses frequency reuse-1 (all cells use the same frequencies), which means every neighboring cell is a potential interferer. The SINR at any point is determined by the ratio of desired signal to the sum of all interfering signals plus noise. At the cell edge, where desired signal is weakest and interference is strongest, SINR can drop below 0 dB.

9.2 ICIC and eICIC

9.3 LTE PCI Planning

The Physical Cell Identity (PCI) is the most fundamental cell-level parameter in LTE. It ranges from 0 to 503 (504 total), composed of two components defined in 3GPP TS 36.211:

9.3.1 Collision & Confusion Rules

Two fundamental constraints must never be violated:

• No collision: Neighboring cells must not share the same PCI. If two adjacent cells have the same PCI, the UE cannot distinguish them during cell search, leading to attach failures and wrong-cell camping.
• No confusion: Two neighbors of the same serving cell must not share the same PCI. If cell A has neighbors B and C both with PCI 50, the eNB cannot determine whether a measurement report refers to B or C — causing handovers to the wrong target cell.

9.3.2 The Three Modular Rules: Mod-3, Mod-6, Mod-30

LTE PCI planning requires satisfying three modular constraints, each arising from how the PCI determines physical signal properties:

9.3.3 Site-Level PCI Assignment

The standard practice assigns PCIs in groups of 3 per site. The consecutive group approach (0,1,2 for site A; 3,4,5 for site B; etc.) automatically satisfies mod-3 but may not optimize mod-6:

9.3.4 PCI Planning Checklist

9.4 PRACH Planning

PRACH (Physical Random Access Channel) is the channel used by UEs to initiate connection with the eNB. PRACH planning is critical for random access success rate (RASR) and affects initial attach time, handover success, and call setup delay.

9.4.1 PRACH Structure & Root Sequences

LTE PRACH uses Zadoff-Chu (ZC) sequences of length 839 (format 0–3, unrestricted) or 139 (format 4, short). Each cell is assigned a root sequence index (rootSequenceIndex, 0–837), from which 64 preambles are generated via cyclic shifts:

9.4.2 Cell Radius & PRACH Root Consumption

The relationship between cell radius and root sequence consumption is the key trade-off in PRACH planning:

9.4.3 PRACH Planning Rules

• Non-overlapping root ranges: Adjacent cells must use different root sequence indices. If cell A uses roots 0–2 and cell B uses roots 0–2, a preamble from a UE at the cell edge may be detected by both → phantom RACH at the wrong eNB.
• Root sequence spacing: Leave guard roots between adjacent cells. If a cell consumes 3 roots, assign cell A roots 0–2, cell B roots 5–7 (gap of 2 roots).
• PRACH configuration index: Determines which subframes carry PRACH. In FDD, configuration 0–15 map PRACH to specific subframes (e.g., config 0 = SFN mod 2 = 0, subframe 1). Adjacent cells can use the same config index but must have different root sequences.
• High-speed flag: For cells covering highways or railways, enable the restricted set (highSpeedFlag = TRUE) which uses a subset of cyclic shifts immune to Doppler shifts. This consumes more roots.
• Preamble group sizing: 64 preambles are split into Group A and Group B. Group B is for UEs with larger message sizes. Sizing matters for capacity: too few Group A → contention; too few Group B → msg3 failure.

10.1 Reference Signals

LTE uses several reference signals for channel estimation, synchronization, and measurement:

10.2 EARFCN & Frequency Calculation

10.3 Timing Advance & Cell Radius

The Timing Advance (TA) compensates for the round-trip propagation delay. The maximum cell radius is determined by the TA range:

10.4 Uplink Power Control — Deep Dive

UL power control is one of the most critical parameters for network optimization. It determines the trade-off between cell-edge coverage (UE needs more power) and inter-cell interference (too much UE power degrades neighbors). Mastering power control tuning is what separates a good RF planner from an average one.

10.4.1 PUSCH Power Control Formula (3GPP TS 36.213 §5.1.1.1)

10.4.2 Open-Loop vs. Closed-Loop Power Control

10.4.3 Power Control Parameter Tuning Guide

10.4.4 NR Power Control Differences

NR extends LTE power control with additional flexibility for beam-based operation:

• Multiple P₀/α sets: NR supports up to 32 P₀/α configurations per BWP (vs. 1 in LTE). Different sets can be activated for different beam directions or service types via DCI.
• Path loss reference: NR allows PL to be measured from SSB or CSI-RS (LTE uses CRS only). In mMIMO, CSI-RS-based PL reflects the actual beam path loss, not the average cell PL.
• SRI-based power control: NR can link power control to the SRS Resource Indicator, enabling beam-specific UL power settings.
• TDD UL power: In TDD, UL slots are shared across all UEs. Power control must balance per-UE fairness vs total UL interference within each slot.

11.1 Key Differences from LTE Link Budget

11.2 NR DL Link Budget — FR1 (C-Band 3.5 GHz)

11.3 NR DL Link Budget — Complete Waterfall (3GPP TS 38.104)

A complete NR DL link budget follows the standard waterfall structure. Each parameter is referenced to its 3GPP specification:

11.4 NR UL Link Budget — The Coverage-Limiting Direction

In NR TDD, the uplink is the coverage bottleneck. While the DL benefits from massive MIMO beamforming (+8 to +12 dB) and high gNB Tx power (49 dBm), the UL is limited by:

• UE max Tx power: 23 dBm (Power Class 3) or 26 dBm (Power Class 2) — cannot match gNB EIRP
• No UE beamforming (FR1): UE has omnidirectional or low-gain antenna → 0 dBi gain in FR1
• UL Rx beamforming at gNB: gNB can apply Rx beamforming (+8 dB gain) to partially compensate

11.5 SCS Impact on Noise Bandwidth

NR’s flexible numerology means the noise bandwidth changes with SCS, directly affecting receiver sensitivity and link budget:

11.6 Beamforming Gain Explained

Beamforming gain is the single most impactful new parameter in NR link budgets. It represents the additional signal strength obtained by coherently combining signals from multiple antenna elements:

11.7 Supplementary Uplink (SUL)

SUL (3GPP TS 38.101, Rel-15) is the primary solution for the DL-UL imbalance at C-band. It allows the UE to transmit on a low-band carrier while receiving on C-band:

• How it works: The gNB configures an SUL carrier (e.g., n28 at 700 MHz) alongside the primary NR carrier (n78 at 3.5 GHz). The UE uses the SUL for PUSCH/PUCCH when UL path loss exceeds a threshold.
• Coverage benefit: Path loss at 700 MHz is ~14 dB less than at 3.5 GHz for the same distance (20×log₁₀(3500/700) = 14 dB). This effectively extends UL coverage to match or exceed DL.
• SUL-aware RACH: SUL has its own PRACH configuration. The UE selects SUL PRACH when measured RSRP from the NR cell is below rsrp-ThresholdSSB-SUL (e.g., -100 dBm).
• Capacity trade-off: SUL bandwidth is typically 5–20 MHz (limited low-band spectrum) vs 100 MHz on C-band. SUL helps coverage but does not add capacity.
• Common SUL band pairs: n78+n28 (3.5 GHz + 700 MHz), n78+n8 (3.5 GHz + 900 MHz), n77+n20 (3.5 GHz + 800 MHz)

11.8 Multi-Band NR Link Budget Summary

12.1 NR Throughput Formula

12.2 Practical NR Cell Throughput

12.3 TDD Slot Configuration Impact

12.4 Massive MIMO Capacity Gain

Massive MIMO (mMIMO) provides capacity gains through two mechanisms: beamforming gain (concentrating energy toward users, improving SINR) and spatial multiplexing (serving multiple users simultaneously on the same time-frequency resource via MU-MIMO). A 64T64R system can typically serve 8–16 users simultaneously with 4–8 spatial layers, delivering 3–5x cell capacity vs. 4T4R.

13.1 SSB-Based Coverage

In 5G NR, initial coverage is determined by the SS/PBCH Block (SSB) transmitted in multiple beam directions during the SS Burst Set. Each SSB beam covers a portion of the cell area. The total cell coverage is the union of all SSB beam footprints.

• FR1 (<6 GHz): Up to 8 SSB beams per cell (L_max=8). Typically 4–8 beams covering the full sector.
• FR2 (mmWave): Up to 64 SSB beams per cell (L_max=64). Narrow beams sweep across wide azimuth/elevation range.
• SSB periodicity: Default 20 ms (can be 5, 10, 20, 40, 80, 160 ms). More frequent = faster cell acquisition but more overhead.

13.2 NR Coverage Thresholds

13.3 Beam Management Procedures

3GPP defines three beam management procedures for maintaining coverage as the UE moves:

• P1 (Beam Sweeping): gNB sweeps SSB beams across the cell; UE measures and reports best beam index. Used for initial access and coarse beam selection.
• P2 (gNB Beam Refinement): gNB transmits CSI-RS in a narrower set of directions around the current beam to refine the DL beam. Triggered when SINR drops.
• P3 (UE Beam Refinement): UE sweeps its receive beams while gNB holds the transmit beam fixed. Used to optimize UE-side beam selection, critical for FR2.

13.4 Coverage Prediction Methodology

NR coverage prediction differs from LTE because of SSB beam sweeping and beamforming gain. The RF planner must account for beam-specific coverage rather than cell-wide omnidirectional coverage:

13.4.1 SS-RSRP Prediction Formula

13.4.2 Coverage Probability Calculation

Coverage probability at a given distance is the probability that the received SS-RSRP exceeds the minimum threshold. With log-normal shadow fading:

13.4.3 Area Coverage Probability

The area coverage probability (percentage of the cell area with SS-RSRP above threshold) is what operators actually target. It differs from edge probability:

13.5 Multi-Band Coverage Comparison

Different NR bands provide vastly different coverage footprints. The RF planner must select the right band for each coverage objective:

13.6 Coverage Planning Workflow

13.7 Co-Planning with LTE (DSS & NSA)

In Non-Standalone (NSA) deployments, 5G NR coverage is anchored by LTE. The UE connects to LTE first, then adds an NR leg via EN-DC. This means NR coverage planning must consider LTE anchor coverage as a prerequisite. Areas with LTE coverage but no NR coverage will have no 5G service even if NR is deployed nearby.

• NSA (EN-DC): NR cell is added as SCG. LTE provides coverage anchor, NR provides capacity boost. NR coverage can be “islands” within LTE coverage.
• SA (Standalone): NR must provide continuous coverage by itself. Coverage gaps cause service drops. More demanding planning than NSA.
• DSS (Dynamic Spectrum Sharing): LTE and NR share same carrier. NR coverage = LTE coverage. Capacity split dynamically based on traffic. Coverage planning is unified.

13.8 Coverage Planning Best Practices

14.1 Active Antenna System Architecture

A massive MIMO Active Antenna System (AAS) integrates the radio unit directly with the antenna panel, eliminating the feeder cable. A typical 64T64R AAS contains 192 antenna elements arranged in a planar array (e.g., 12 columns × 8 rows of cross-polarized elements). Each pair of elements is driven by a dedicated transceiver chain with independent phase and amplitude control.

14.2 Vendor AAS Hardware Comparison

Selecting the right AAS product is one of the most impactful decisions in 5G planning. Each vendor offers different element counts, weights, power consumption, and beam capabilities:

14.3 Beam Pattern Modeling

Understanding how the AAS forms beams is essential for coverage prediction. The beam pattern depends on element count, element spacing, and the applied beamforming weights:

14.3.1 Element Spacing & Array Factor

14.3.2 Sidelobe Control

When forming narrow beams, sidelobes appear at angles away from the main beam. Sidelobes cause interference to neighboring cells and users on other beams:

• First sidelobe level: For a uniform linear array without tapering, the first sidelobe is ~13 dB below the main lobe. With amplitude tapering (e.g., Taylor window), sidelobes can be reduced to -20 to -25 dB.
• Element spacing: Standard λ/2 spacing (43 mm at 3.5 GHz) avoids grating lobes. If spacing exceeds λ/2, grating lobes appear at large scan angles, causing interference in unintended directions.
• Planning impact: Sidelobes from one cell can interfere with users in adjacent cells. Higher sidelobe suppression improves cell-edge SINR but reduces peak beamforming gain slightly. Vendor-specific implementations vary.

14.4 CSI Feedback & Codebook Design

Beamforming accuracy depends on channel state information (CSI) fed back from the UE to the gNB. NR supports two CSI frameworks:

14.5 MU-MIMO Capacity Gain

The primary capacity benefit of massive MIMO is multi-user MIMO (MU-MIMO) — serving multiple UEs simultaneously on the same time-frequency resources using spatial separation:

14.6 Practical mMIMO Deployment Planning

15.1 NR PCI Planning

Physical Cell Identity (PCI) planning is one of the most critical tasks in 5G NR network design. Incorrect PCI assignment causes cell search failures, measurement ambiguity, handover problems, and degraded throughput. NR doubles the PCI space compared to LTE and introduces new modular constraints tied to SSB, PBCH DMRS, and CSI-RS. This section provides a comprehensive guide to NR PCI planning.

15.1.1 NR PCI Structure & Range

NR PCI ranges from 0 to 1007 (1008 total), exactly double LTE’s 504. Each PCI is composed of two components defined in 3GPP TS 38.211 Section 7.4.2.1:

The UE performs cell search in two stages: first detecting N_ID⁽²⁾ from the PSS (3 hypotheses), then detecting N_ID⁽¹⁾ from the SSS (336 hypotheses). This two-stage design keeps cell search complexity manageable even with 1008 PCIs.

15.1.2 PCI Collision & Confusion

Two fundamental constraints govern PCI assignment in any cellular network. Violating either causes serious degradation:

15.1.3 NR Modular Rules: Mod-3, Mod-4, Mod-8

Beyond collision and confusion avoidance, NR PCI planning must respect three modular constraints. Each arises because different physical channels and signals derive their sequence or resource position from the PCI value:

15.1.4 Mod-3 Rule — PSS Sequence (Most Critical)

The mod-3 rule is the most important PCI planning constraint, shared between LTE and NR. Since N_ID⁽²⁾ = PCI mod 3 directly determines the PSS sequence, two co-located sectors with the same mod-3 value transmit identical PSS sequences. The UE performing initial cell search cannot distinguish them, causing:

• Delayed cell search: UE takes 2–5× longer to lock onto the correct cell
• Wrong cell camping: UE may camp on a weaker cell and experience poor throughput
• Increased RRC setup failures: Cell barring due to PSS/SSS ambiguity
• Handover ping-pong: UE oscillates between cells it cannot properly distinguish

15.1.5 Mod-4 Rule — PBCH DMRS (NR-Specific)

This rule is unique to NR and has no LTE equivalent. In NR, the PBCH carries the Master Information Block (MIB), which the UE must decode to access the cell. The DMRS sequence for PBCH is initialized based on PCI mod 4 (TS 38.211 Section 7.4.1.4.1):

• 4 possible PBCH DMRS initialization states based on PCI mod 4
• If two neighboring cells share the same PCI mod 4, their PBCH DMRS sequences are correlated
• At cell edge where both SSBs are received at similar power, the UE struggles to decode MIB
• Effect: Increased PBCH BLER → delayed system acquisition → poor initial access experience
• Planning rule: Adjacent cells and co-sector cells should have different PCI mod 4 values

Since mod-4 provides only 4 values {0, 1, 2, 3}, it is impossible to guarantee unique mod-4 across all neighbors in dense deployments. Prioritize: (1) co-sector diversity, (2) strongest first-tier neighbors.

15.1.6 Mod-8 Rule — CSI-RS (NR-Specific)

CSI-RS (Channel State Information Reference Signal) is the primary reference signal for channel estimation in NR (replacing LTE’s CRS). The CSI-RS resource element mapping depends on PCI mod 8, creating 8 possible RE shift patterns:

• Two cells with the same PCI mod 8 transmit CSI-RS on the same resource elements
• At cell edge, CSI-RS from both cells collide → corrupted channel estimation
• Effect: Incorrect CQI/PMI/RI reports → wrong MCS selection → high BLER → throughput degradation of 15–30% at cell edge
• This is especially critical for massive MIMO where accurate CSI is essential for beamforming weights
• Planning rule: Maximize mod-8 diversity among cells within the beamforming coordination area

15.1.7 Practical Site-Level PCI Assignment

The standard practice is to assign PCIs in groups of 3 (one group per site), ensuring each sector gets a different mod-3 value. Here is the recommended approach:

For large-scale networks, PCI planning follows a graph coloring approach:

• Step 1: Build a neighbor graph where each cell is a node and edges connect cells with overlapping coverage
• Step 2: Assign PCI groups to sites using graph coloring algorithms (greedy or constraint satisfaction) to minimize mod-3 conflicts among neighbors
• Step 3: Within each group of 3, assign mod-3 values {0,1,2} to the 3 sectors
• Step 4: Verify mod-4 and mod-8 diversity among first-tier neighbors — swap groups if needed
• Step 5: Validate with collision/confusion audit tool (automated scanners or vendor OSS)

15.1.8 NR PCI Planning — Complete Checklist

15.1.9 SA vs. NSA PCI Planning Considerations

PCI planning differs depending on the deployment mode:

15.1.10 Common PCI Planning Mistakes

15.2 SSB Frequency Position (GSCN)

The Synchronization Signal Block (SSB) is the most important broadcast signal in NR. It carries the PSS, SSS, and PBCH, and is the entry point for every UE performing cell search. The SSB’s frequency position must be placed on the synchronization raster — defined by the Global Synchronization Channel Number (GSCN) — so that UEs can find it efficiently.

15.2.1 Why GSCN Exists

Unlike LTE where the PSS/SSS are always at the center of the channel bandwidth, NR decouples the SSB position from the channel center frequency. This enables:

• Flexible SSB placement: SSB can be offset from center, allowing asymmetric BWP configurations
• Faster cell search: The GSCN raster is coarser than the channel raster (100 kHz), reducing the number of frequency hypotheses the UE must scan
• Shared spectrum: Multiple operators sharing a band can place SSBs at different GSCN positions to avoid SSB collision

15.2.2 GSCN Formulas

3GPP TS 38.104 Table 5.4.3.1-1 defines three GSCN-to-frequency formulas depending on the frequency range:

15.2.3 GSCN for Common NR Bands

15.2.4 SSB Planning Considerations

• SSB within active BWP: The SSB must fall within the initial DL BWP. If the BWP is narrower than the channel BW, verify the selected GSCN places the SSB inside the BWP.
• Avoid SSB at band edge: Placing SSB near the edge of the carrier causes guard band issues and may degrade PSS/SSS correlation. Place SSB near center for best performance.
• Multi-operator coordination: In shared spectrum (n77/n78), operators must coordinate GSCN to avoid SSB overlap. Different GSCN positions = different SSB frequencies = no SSB collision.
• SSB periodicity: Configurable at 5, 10, 20, 40, 80, 160 ms. Default during initial cell search: 20 ms. Shorter periodicity = faster measurements but more overhead.
• SSB beam sweeping: For FR1 with SCS 30 kHz, up to 8 SSB beams (L_max=8). For FR2 with SCS 120 kHz, up to 64 SSB beams. Each beam carries the same GSCN frequency — only the spatial direction changes.

15.2.5 SSB Configuration — Number of Beams (L_max)

15.3 TDD Frame Synchronization

TDD frame synchronization is the single most critical deployment requirement for 5G NR TDD networks. Unlike FDD where uplink and downlink use separate frequencies, TDD shares one frequency band and separates DL and UL in the time domain. If neighboring cells are not synchronized, catastrophic cross-link interference (CLI) occurs.

15.3.1 The Cross-Link Interference Problem

CLI happens when one cell transmits downlink while a neighbor is receiving uplink on the same frequency at the same time:

15.3.2 Synchronization Requirements

TDD synchronization has three dimensions that must all be aligned across neighboring cells:

15.3.3 Common TDD Slot Patterns

15.3.4 GPS/GNSS Timing & Guard Period

• Timing source: All TDD cells must be phase-synchronized to a common clock (GPS/GNSS, IEEE 1588v2 PTP, or SyncE). Accuracy requirement: ±1.5 μs (3GPP TS 38.104).
• Guard period (GP): The special slot contains a guard period between DL and UL symbols. GP duration must accommodate the round-trip propagation delay from the farthest UE: GP ≥ 2 × R_max/c. For 5 km cell radius: GP ≥ 33 μs ≈ 1 OFDM symbol at 30 kHz SCS.
• Timing advance: In TDD, the TA correction applies to UL timing. The gNB commands TA so that UL arrives at the correct symbol boundary. Incorrect TA causes UL-to-DL transition overlap.
• Multi-operator sync: On shared TDD bands (n77, n78), regulators mandate synchronized TDD patterns across all operators. Cross-operator CLI is the main risk if operators use different DL/UL ratios.

15.4 PRACH Planning for NR

NR PRACH planning is significantly more complex than LTE due to flexible numerology, multiple preamble formats, and the interaction with SSB beams. Proper PRACH planning directly impacts RACH success rate (RASR), initial access latency, and handover performance.

15.4.1 NR PRACH Preamble Formats

NR defines two categories of PRACH preamble formats, each serving different deployment scenarios:

15.4.2 PRACH-SSB Association

A critical NR-specific concept is the PRACH occasion-to-SSB beam mapping. Each PRACH occasion is associated with one or more SSB beams, ensuring the gNB knows which beam the UE detected before initiating RACH:

• SSB-per-RACH-occasion: Configures how many SSB beams share a single PRACH time/frequency occasion. Values: 1/8, 1/4, 1/2, 1, 2, 4, 8, 16. Lower ratios = more PRACH resources needed.
• Contention-based (CB) preambles: 64 per cell (same as LTE), split across SSB beams. With 8 SSB beams and 64 preambles, each beam gets only 8 preambles → higher contention probability.
• Contention-free (CF) preambles: Dedicated preambles assigned per UE for handover and beam failure recovery. Reduces CB pool further.
• msg1-FDM: NR supports 1, 2, 4, or 8 PRACH occasions in frequency domain per time slot, increasing RACH capacity.

15.4.3 NR PRACH Root Sequence Planning

NR root sequence planning follows the same principle as LTE — adjacent cells must have non-overlapping root indices — but with added complexity from two sequence lengths:

• Long sequences (L=839): Same ZC-based approach as LTE. Root sequence index 0–137 (restricted set) or 0–837 (unrestricted). Cyclic shift configuration via zeroCorrelationZoneConfig (0–15).
• Short sequences (L=139): Used with short PRACH formats (A1–C2). Root index 0–137. Fewer cyclic shifts available per root → more roots consumed per cell. Typical: 2–8 roots per cell for urban deployments.
• FR2 PRACH: Uses short sequences with 60/120 kHz SCS. Very small cells (<500m) → 1–2 roots per cell, but beam-sweeping RACH adds complexity.

15.4.4 NR PRACH Planning Checklist

16.1 mmWave Propagation Characteristics

Millimeter wave frequencies (24.25–52.6 GHz in FR2, extendable to 71 GHz in FR2-2) offer enormous bandwidths (100–400 MHz per carrier) but suffer from propagation losses that are fundamentally different from sub-6 GHz. Understanding these losses is the foundation of all mmWave planning.

16.1.1 Loss Mechanisms

16.1.2 Propagation Models for FR2

3GPP TR 38.901 defines path loss models for FR2 with LoS and NLoS scenarios:

16.1.3 Rain Attenuation Calculation

ITU-R P.838 provides specific rain attenuation coefficients for mmWave planning. Rain margin must be included in the link budget:

16.2 mmWave Link Budget

The mmWave link budget follows the same structure as FR1 but with critical differences: massive beamforming gain, atmospheric/rain losses, and dramatically shorter cell range.

16.3 FR2 Beam Management

Beam management is the defining feature of mmWave operation. With narrow beams (5–10° beamwidth), both the gNB and UE must continuously align their beams. 3GPP defines a three-level beam management procedure:

16.4 mmWave PCI Planning

PCI planning for FR2 follows the same mod-3/mod-4/mod-8 rules as FR1, but with critical differences due to the much larger number of SSB beams:

16.4.1 64 SSB Beams and PCI Implications

• L_max = 64: FR2 cells can transmit up to 64 SSB beams per burst set. Each beam covers a narrow angular sector (~5°). All 64 beams share the same PCI.
• Mod-3 rule: Still critical — all 64 beams of a cell share the same N_ID⁽²⁾. Co-site cells must have different mod-3 values.
• Mod-4 rule: Each of the 64 SSBs carries PBCH with DMRS sequence determined by PCI mod 4. Neighbor cells with same mod-4 cause PBCH decode issues on all 64 beams simultaneously.
• Mod-8 rule: CSI-RS patterns used for beam refinement (P2/P3) depend on PCI mod 8. In dense mmWave deployments with many small cells, mod-8 diversity is especially important.

16.4.2 Dense Deployment Challenges

16.5 mmWave PRACH Planning

PRACH planning for FR2 is fundamentally different from FR1. The combination of 64 SSB beams, short preamble formats, beam-swept random access, and dense small cell deployments creates a multi-dimensional optimization problem that directly impacts initial access latency, handover success rate, and beam failure recovery time.

16.5.1 FR2 PRACH Preamble Formats — Detailed

FR2 exclusively uses short preamble sequences (L_RA = 139) with subcarrier spacings of 60 kHz or 120 kHz. Long preambles (L=839) are never used in FR2 because the cell radius is too small to need them, and their 1.25/5 kHz SCS doesn’t align with FR2 numerology.

16.5.2 Beam-Swept RACH — The 64-Beam Challenge

The defining challenge of FR2 PRACH is mapping 64 SSB beams to PRACH occasions. Every SSB beam must have at least one associated PRACH occasion where a UE can transmit its preamble. This creates a resource dimensioning problem unique to mmWave:

16.5.3 Contention-Based Preamble Collision Analysis

With fewer preambles per SSB beam in FR2, contention-based preamble collisions become a real concern. The collision probability depends on the number of UEs attempting RACH simultaneously on the same beam:

16.5.4 FR2 4-Step RACH Procedure

The FR2 RACH procedure follows the standard NR 4-step RACH but with beam-specific enhancements:

16.5.5 Power Ramping & Beam-Specific Considerations

FR2 PRACH power control has unique aspects due to beamforming at both the gNB and UE:

• Beam correspondence: FR2 UEs are required to support beam correspondence (3GPP TS 38.101-2), meaning the UE Tx beam direction aligns with the best Rx beam direction. This eliminates the need for separate UL beam sweeping during RACH.
• Power ramping across beams: If the UE fails RACH on the best SSB beam (no RAR received within ra-ResponseWindow), it may switch to the next-best SSB beam before continuing power ramping. This is configured via ssb-perRACH and the RACH resource selection procedure.
• Maximum retransmissions: preambleTransMax (typically 7 or 10 for FR2). After exhausting retransmissions, UE declares RACH failure and may trigger beam failure recovery or cell reselection.
• Power ramping step: 2 dB for FR2 (vs. 2–4 dB for FR1). Smaller steps because mmWave path loss is more variable and aggressive ramping wastes UL resources.

16.5.6 FR2 Root Sequence Planning

Root sequence planning for FR2 is simpler than FR1 due to small cell sizes:

16.5.7 Complete FR2 PRACH Planning Checklist

16.6 Deployment Strategies

mmWave deployments follow fundamentally different strategies than sub-6 GHz. The key principle: plan for LoS coverage, not area coverage.

16.6.1 Deployment Scenarios

16.6.2 Site Selection Criteria

• LoS analysis is mandatory: Use 3D building models or LiDAR data to verify LoS from the candidate site to the target coverage area. RF predictions alone are insufficient for mmWave.
• Avoid tree-lined streets: Even seasonal foliage can add 20–40 dB loss. Prefer sites that look down corridors free of vegetation.
• Street-facing orientation: Mount antennas facing along streets, not perpendicular to building facades. Street canyon reflections extend NLoS coverage.
• Height optimization: Too high = large shadow zone behind buildings. Too low = limited LoS range. Optimal height depends on street geometry — typically 6–12 m for street-level.
• Glass facades: Modern buildings with Low-E coated glass block mmWave completely. Do not plan outdoor-to-indoor coverage for mmWave.

16.7 Integrated Access & Backhaul (IAB)

IAB (3GPP Rel-16) allows mmWave small cells to use the same mmWave spectrum for both user access and backhaul connection to the core network. This eliminates the need for fiber to every small cell — a game-changer for dense mmWave deployment economics.

16.7.1 IAB Architecture

• IAB Donor: Fiber-connected gNB that serves as the anchor point. Provides both access to UEs and backhaul to IAB nodes.
• IAB Node: Wireless relay that connects to the IAB donor (or another IAB node) via the backhaul link, and serves UEs via the access link.
• Multi-hop: IAB supports up to 2–3 hops (IAB Donor → IAB Node 1 → IAB Node 2 → UE). Each hop adds latency (~1–2 ms) and reduces available capacity.
• In-band vs out-of-band: In-band IAB uses same mmWave frequency for access and backhaul (time-multiplexed). Out-of-band uses different carriers (higher capacity but more spectrum).
• Topology management: The IAB donor CU manages the routing topology. IAB nodes can switch parent nodes for resilience (Rel-17 enhancement).

16.8 mmWave Capacity Planning

mmWave delivers massive capacity per cell due to wide bandwidths, but the small cell size means capacity planning focuses on throughput per cell and cells per area:

16.9 Fixed Wireless Access (FWA) Planning

FWA is the strongest near-term business case for mmWave. It provides fiber-like broadband to homes/offices using a directional outdoor CPE:

• CPE antenna gain: +20 to +25 dBi directional antenna mounted on building exterior or window. This dramatically extends range compared to handheld UE.
• FWA link budget: With CPE gain, effective MAPL increases by 10–15 dB → LoS range extends to 500–1500 m (vs. 200 m for smartphone).
• LoS requirement: CPE must have clear LoS to the gNB. Professional installation required to aim the CPE antenna. Self-install kits use wider beams but shorter range.
• Throughput per CPE: 300 Mbps–1 Gbps typical (dedicated bandwidth, not shared). Comparable to fiber for most residential use cases.
• Subscribers per sector: 30–100 CPEs per mmWave sector with 2 Gbps cell throughput (depending on overbooking ratio).
• Rain margin: FWA links are longer than mobile → need 3–5 dB rain margin even in temperate climates.

16.10 mmWave Planning Checklist

17.1 Why Indoor Planning Matters

Approximately 80% of mobile data traffic originates indoors, yet indoor coverage from outdoor macro cells is increasingly challenging due to modern building materials (Low-E glass: 25–40 dB loss). Dedicated indoor solutions are now essential for quality service, especially at C-Band and above.

17.2 Indoor Propagation (ITU-R P.1238)

17.3 Indoor Solution Types

18.1 Dimensioning Workflow

Network dimensioning is the first quantitative step in RF planning. It answers the fundamental question: How many sites do we need? The workflow follows these steps:

• Step 1 — Traffic Forecast: Estimate subscribers, traffic per user (GB/month), busy hour ratio, DL/UL split, and growth rate over 3–5 years.
• Step 2 — Coverage Dimensioning: Calculate MAPL from link budget → cell radius from propagation model → site count = Area / (cell coverage area × sectors).
• Step 3 — Capacity Dimensioning: Calculate required throughput per area → divide by cell throughput → site count for capacity.
• Step 4 — Take Maximum: Final site count = max(coverage sites, capacity sites) per morphology.
• Step 5 — Backhaul & Core: Dimension transport (fiber, microwave) and core network elements.

18.2 Coverage-Based Site Count

18.3 Backhaul Dimensioning

Each cell site requires backhaul capacity proportional to its peak throughput. For a 5G site with 3 sectors of 100 MHz C-Band (64T64R), peak aggregated throughput can reach 10+ Gbps, requiring fiber or high-capacity microwave backhaul (E-Band 70/80 GHz). LTE sites typically need 200–500 Mbps per site.

19.1 Tower Types

• Lattice/Self-Support: Steel lattice structure, 30–100 m. Highest capacity for equipment. Common in rural areas.
• Monopole: Single steel tube, 15–40 m. Smaller footprint, aesthetically better. Common in suburban/urban.
• Rooftop: Antenna mounted on building roof using non-penetrating mounts or parapet brackets. 3–6 m above roof. Most common in dense urban.
• Concealed/Camouflaged: Disguised as trees (monopine), flag poles, chimneys, or building features. Required in aesthetically sensitive areas.
• COW (Cell on Wheels): Temporary mobile tower on trailer. Used for events, emergencies, and gap coverage during construction.
• Street Furniture: Small cells mounted on lamp posts, bus shelters, or utility poles. 5–10 m height. Used for 5G densification.

19.2 Antenna Mounting Guidelines

• Mount antenna above local clutter with minimum 2 m clearance from any obstruction
• Maintain minimum 1.5 m vertical separation between antenna bands to avoid intermodulation (PIM)
• Ensure antenna connector torque per manufacturer spec (typically 15–25 Nm)
• Use weatherproof tape and boots on all outdoor connectors
• Route cables in cable trays, secured every 1 m, with drip loops at antenna entry
• Label all cables at both ends with sector, band, and port information

19.3 Site Survey Checklist

Before finalizing a site, an RF engineer must conduct a physical site survey to verify suitability. Key checklist items include: GPS coordinates, photos (360° panoramic), antenna mounting positions, height above ground, clear line-of-sight check, structural assessment, power availability, backhaul readiness, access road condition, and landlord/permitting status.

20.1 Drive Test Equipment

• UE/Scanner: Rohde & Schwarz TSMA, Keysight Nemo Walker/Outdoor, or commercial smartphones with measurement apps (Nemo Handy, XCAL).
• GPS: High-accuracy GPS receiver (sub-3 m accuracy) for precise location tagging.
• Software: Nemo Outdoor, XCAP, Actix, TEMS for data collection. Post-processing with Nemo Analyze, Actix Analyzer, or custom Python/MATLAB scripts.
• Vehicle: Roof-mounted GPS and measurement antennas. UE positioned at consistent height (~1 m) inside vehicle.

20.2 Drive Test KPIs

20.3 Model Calibration Process

After collecting drive test data, the propagation model is calibrated by adjusting K-factors to minimize the difference between predicted and measured path loss. The target is RMSE < 8 dB for macro cells (6 dB for excellent calibration). Separate calibrations are performed for each morphology class (dense urban, urban, suburban, rural).

21.1 The Optimization Lifecycle

RF optimization is not a one-time activity — it is a continuous iterative loop that runs throughout the network’s lifetime. Every new site, traffic pattern change, or subscriber growth triggers re-optimization.

21.2 KPI Framework & Targets

Every optimization activity is driven by KPIs. The RF optimizer must monitor these KPIs continuously and trigger corrective action when any falls below target: