Addressing Anomalous Interference

within Mobile Networks

www.spectrumeffect.com
June 4, 2020
© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 1
Spectrum Effect®
Pioneering Spectrum Analytics
❖ Founded in 2015 by wireless industry
veterans behind Eden-NET SON
(acquired by Nokia).
❖ Innovative technologies with roots in
spectrum sharing programs with the
United States DoD and ONR.
❖ R&D in Kirkland, WA USA and Monterrey,
Mexico.
❖ 20 issued patents and 20+ pending
patent applications.
❖ Gartner 2019 Cool Vendor.
$1T USD*
❖ Revolutionary spectrum analytics solution VALUE OF SPECTRUM
built on machine learning framework:
GLOBALLY
Spectrum-NET.
*Source: extrapolated from 2019 earnings reports from leading mobile operators

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 2
Spectrum-NET®
Machine Intelligence Meets Spectrum
❖ Automated, seamless RF interference
management across multi-vendor UMTS, LTE
and (soon) 5G NR networks.
❖ Innovative ML-based algorithms applied to
operator RAN data.
❖ Detection, characterization, classification,
aggregation, locating, prediction and impact
assessment of anomalous RF interference.
❖ Web-based, cloud-ready, containerized,
highly scalable solution with extensive
Anomalous RF Interference Categories
mapping, visualization, and reporting
capabilities. External Interference Cross Border Interference
BDA’s, CATV Egress, RFID Scanners, Broadcast
Continuous transmissions.
❖ Completed first trial with 5G NR network in Q4 TV, RF Jammers, Grey Market Devices, Legacy
Equipment, …
Intermittent during periods of tropospheric ducting.
2019.
TDD Self-Interference RAN Generated Interference
Typically occurs during periods of tropospheric From UE traffic.
ducting. Requires Network Optimization.
PIM Interference Faulty Hardware
Internal and external sources at the site. Issues with radio units, cabling or antenna.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 3
Spectrum-NET
Advanced Interference Analysis
Data Ingestion
❖ RAN CM, PM and topology data are
ingested by Spectrum-NET.
Advanced Interference Analysis
❖ 7 analysis stages.
❖ Captures comprehensive actionable
event info for expeditious resolution.
❖ Prioritizes events via KPI impact,
interference type, network scope,
etc.

ML to Continually Learn and Refine Results

❖ Includes library of classification models learned from many networks.
❖ Rules are refined and new rules are created via unsupervised, supervised and deep machine learning.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 4
Spectrum-NET
Interference Characterization, Aggregation and Locating

❖ Spectrum-NET characterizes the interference by time, frequency, power and

per-ROP activity %.
❖ Spectrum-NET aggregates all the cells that are being impacted by
interference with a common characteristic.
Validated extensively in operator
❖ Spectrum-NET calculates the interference magnitudes at each cell and networks across the globe.
applies weighted Bayesian inference to calculate accurate locating of the
interference. Algorithm applies both horizontal and vertical antenna radiation Example Interference Event from
pattern information and Shuttle Radar Topography Mission data. a BDA in San Diego.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 5
Spectrum-NET
External RF Interference Examples
Bidirectional Amplifiers (Seattle, Toronto, Jacksonville, Hanoi and Tunis)

Terrestrial Broadcast TV (San Francisco) RF Jammers (Mexico City, Nassriya and Cebu)

WiMAX Interference (Buenos Aires, Mexico City and Puerto Rico)

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 6
Spectrum-NET
External RF Interference Examples
DECT Devices (Las Vegas and San Francisco ) CATV Cable Egress (Buenos Aires and Lima)

Faulty Radio Unit (Abu Dhabi, San Francisco and Tallinn) Faulty DAS (Bangkok) Streetlamps (Puerto Rico)

Defense Facility (Milan) Cross Border (USA to Canada, Mexico to USA, Uruguay to Argentina)

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 7
Spectrum-NET
Passive Intermodulation (PIM) Interference
PIM Interference Spectrum-NET Automatic PIM Interference Detection
❖ Caused by nonlinearities present at different positions ❖ Analyzes site frequencies to generate PIM Spectrograph
along the RF signal’s path. showing where the potential for PIM exists.
❖ “Internal” PIM is from sources between the RF hardware ❖ Deep machine learning has been applied to identify the
and antenna (e.g., degraded connectors and corroded RF signature of PIM within PM data.
cabling).
❖ Automatically identifies this signature for internal and external
❖ “External” PIM is from sources sources beyond the PIM cases (IM2 – IM5).
antenna (e.g., unintended reflection plates).
❖ Broadband interference often impacts the entire channel; KPI
❖ Can impact any band. Potential increases as more bands degradations vary based on interference levels.
are deployed; frequencies from other operators can
❖ Measures the KPI impact PIM and calculates PIM severity score
contribute when site sharing. Interference levels will vary
for prioritizing which sites to address.
depending on the traffic and weather conditions.

PIM Interference Example (Phnom Penh) PIM Interference Example (Bari)

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 8
Spectrum-NET
PIM Interference Event Actionable Reports

Sample report from Spectrum-NET with list of cells Impacted by PIM

interference sorted by KPI Impact and PIM interference severity.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 9
Spectrum-NET
Remote Confirmation of PIM Contributing Frequencies
▪ Once PIM interference is detected at a cell, remote testing
can be performed to confirm the contributing downlink
channels prior to the site visit.
▪ Possible contributing cells are blocked for ~2 PM data
reporting periods so those cells are not transmitting
downlink signals.
▪ The interference levels are measured on the cell with PIM
interference while the downlink channels are blocked.
▪ These measurements are compared with other periods
when channels were not blocked.
▪ If the interference levels are substantially reduced, then the
contributor cells have been confirmed.

Spectrum-NET will be integrated with closed-loop SON: enabling automatic launch, results analysis, and report
generation on the confirmation of PIM interference contributors.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 10
Spectrum-NET
TDD Self-Interference via Tropospheric Ducting
❖ Under ducting weather conditions, signals propagate greater distances and LTE TDD cells in neighboring deployment
regions can cause interference to each other. This will also be a major challenge for 5G NR.
❖ Spectrum-NET dynamically detects victim cells based on common time-frequency signatures. Victim cells can likewise be
identified for cross border interference during and after ducting conditions for both TDD and FDD bands.
❖ Spectrum-NET dynamically detects the aggressor cells (causing the interference) based on interference locating, topology
analysis, traffic correlation and reciprocity analysis.
❖ Upon detection, Spectrum-NET triggers dynamic reconfiguration to mitigate the impact of ducting interference (i.e., SSF is
reconfigured to increase guard periods (GP) at only the Aggressor cells during the ducting).
Aggressor

Self-Interference
Victim in New Delhi

Self-Interference in
Jacksonville

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 11
Spectrum-NET
RAN Interference Management
❖ RAN anomalous interference is mostly present in urban and sub-
urban areas where a high concentration of mobile traffic is present.
❖ RAN interference cause significant performance degradation to the
affected cells.
❖ Spectrum-NET detects cells which are affected by RAN anomalous
interference.
❖ Cells with anomalous levels of RAN interference are clustered in
events based on location and pattern correlation.
❖ To mitigate RAN interference, optimization actions such as load
balancing, down tilting and power control can be triggered.
❖ Spectrum-NET will integrate with C-SON to provide actions to
mitigate high levels of uplink interference.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 12
Spectrum-NET
Interference Event Monitoring
Dashboard View of Anomalous Interference
❖ View of all the interference events affecting the network.
❖ View the distribution of events by time, source, technology
and RF band.
❖ Sort, filter and select events by different criteria including
severity, number of cells, source type, …

Interference Event Details

❖ In-depth view of specific interference events.
❖ Interference location map, frequency characteristics and
time-prediction.
❖ KPI impact of the interference event.

Report Generation
❖ User configurable on-demand and scheduled reporting in
XLS and PPT formats.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 13
 Spectrum-NET Global Dashboard
RF Interference Demystified
Key Results – measured across the networks:
6 ▪ ~5% of Cells Experience Severe Anomalous Interference
7 ▪ Anomalous Interference is active 50% of the time
4
▪ Sources are 65% external, 25% RAN and 10% PIM
1 7 ▪ Typical degradation to network KPI’s from Anomalous
Interference events on the impacted cells includes a
3
200% increase in Drop Rate, 200% increase in Service
Setup Failure Rate, 75% increase in Handover Failure
36%
Rate, 30% increase in Latency, 30% decrease in
52%
LTE
UMTS Downlink/Uplink traffic
FDD
12% LTE TDD
Spectrum-NET Benefits:
▪ Resolve many more interference events with much
greater efficiency.
Operator Impact from Anomalous Interference: ▪ Optimal SSF configuration settings for TDD networks.
▪ Degraded network KPI’s and lost network capacity ▪ Optimize network uplink performance.
▪ Poor subscriber quality of experience ▪ Improve network KPIs.
▪ Elevated trouble tickets ▪ Improve subscriber QoE.
▪ Costly and time consuming to address with current practices ▪ Improve network capacity.
▪ Issues go undetected and unsolved for long periods

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 14
Spectrum-NET
Software Architecture
Data Ingestion
❖ CM, PM, and topology data
RESTful API’s
❖ Interference analysis results
❖ Recommended Mitigation
❖ Network Abstraction Model
Micro-Service Architecture
❖ All Spectrum-NET components are
deployed as micro-services
Container Orchestration
❖ Using Kubernetes container
orchestration framework
❖ Efficient use of VM resources
❖ Horizontally scalable

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 15
 Spectrum-NET Roadmap
Powerful New Capabilities Coming Soon!
Regulator Portal 5G NR Support

New with 5G NR Networks Impact on Interference Environment

Higher Spectral Efficiencies Requires a cleaner RF environment

Integration with Portable SON Integration
Spectrum Analyzers Greater susceptibility to unintended
TDD Band Allocations
Interference (TDD guard times)

Critical new need to identify/locate

mm-Wave Frequency Bands
obstructions

Many incumbent devices will need to be

Mid-band Spectrum
cleared

High quality channel conditions required to

Massive MIMO
realize MIMO gains

Spectrum-NET Becomes Even More Essential for 5G NR Networks

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 16
 Spectrum-NET
Application Level Integration with SON

❖ Spectrum-NET automatically triggers

SON to take actions to in response to SON
interference events.
❖ SON Modules automatically mitigate
specific external interference events:
▪ E.g., dynamically changing SSF of TDD
aggressor cells during ducting.
❖ SON Module perform automated
troubleshooting of interference events
to validate analysis.
❖ Spectrum-NET triggers specific existing SON functionalities based on characteristics of RAN-generated interference.
▪ E.g., change the DL footprint of the cells by modifying idle mode parameters on those cells so idle mode UEs at the edge of the cell don’t
camp in the interfered cells. Only UEs closer to the cell will camp on the cell.
▪ E.g., change handover parameters at neighboring cells to prevent or restrict handovers to cells with interference.

Spectrum-NET automatically triggers actions to minimize the impact of the interference.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 17
Spectrum-NET
2020 Roadmap
Spectrum-NET
Key Feature Feature Description
SW Release
Allows users to annotate interference events with information such as descriptions of
20.07 Interference Event Annotation
interference, searchable labels and photos and videos of interference sources.

20.08 Per-ROP Interference Updates Near-real time updates on the status of interference events at the rate of arrival of PM data.

20.09 Samsung Support Support for Samsung RAN vendor for LTE.

Incorporation of 5G NR counter and CM data into Spectrum-NET and use of this data in
20.10 Support for 5G NR interference analytics and reporting. Support for existing RAN vendors: Huawei, Ericsson,
Nokia and Samsung.
Integration with SON solutions via Spectrum-NET REST API. Enables the development of
SON Modules that automatically mitigate interference events and automation of
20.12 Closed-Loop Actions via SON Integration
interference events troubleshooting to validate analysis. Triggers SON functionalities based
on characteristics of the interference.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 18
Spectrum-NET
2021 Roadmap
Spectrum-NET
Key Feature Feature Description
SW Release
Remote access for regulators (explicitly authorized by the operator) to view external interference
events that operators have submitted to the regulator for action by the regulator. The regulator will
Regulator Portal
have a view of the history, network impact, latest near real-time status, location for the reported
21.Q1 events and can update the events with their actions.

Download of information about interference events from Spectrum-NET to the portable spectrum
Integration with portable spectrum
analyzer. Near real-time interference updates to the spectrum analyzer from Spectrum-NET. Field
analyzer Phase 1
results from the portable spectrum analyzer to Spectrum-NET.

Spectrum-NET will ingest data from ducting prediction maps for enhanced classification of
Prediction of Ducting Conditions
interference resulting from tropospheric ducting.
21.Q2
Multi-Operator PIM Detection Use of data from multiple operators to perform PIM detection across operators with shared sites.

Enhancement of the interference prediction engine using ARIMA and LSTM based Deep Learning
Interference Prediction Phase 2
models to provide more accurate information about the expected interference levels.
21.Q3
Support for cells operating in Dynamic Analysis of anomalous interference events on channels that dynamically switch between LTE and
Spectrum Sharing (DSS) 5G NR.

Identify interference affecting UE Rx frequencies. Uses measurement reports and other cell phone
21.Q4 DL Interference Analysis
generated events to detect and localize DL interference sources.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 19
Spectrum-NET
2022 Roadmap
Spectrum-NET
Key Feature Feature Description
SW Release
Integration with portable spectrum Spectrum-NET interference locating algorithm running on portable spectrum analyzer for updating
analyzer Phase 2 interference location via field measurements.
22.Q1
5G TDD Combined DL/UL Analysis for Combines DL and UL interference analysis including locating for TDD for external interference
External Interference events that span DL and UL timeslots for 5G NR.

Integration with Spectrum Access Integration of Spectrum-NET with the SAS to lookup interference sources within the SAS as
Systems (SAS) for Spectrum Sharing Spectrum-NET locates interference events.
22.Q2
Play back functionality in the UI to show the location of interference events over time. Fast forward
Interference Locating Playback View
and pause functionality will be added as in a movie player.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 20
Spectrum-NET Audit Project
3-Month Audit of the RF Interference Environment
Interference Analysis of all UMTS, LTE and 5G NR Cells within the Selected Markets
1. Spectrum Effect installs and administers Spectrum-NET in the AWS environment for this audit project.
2. Operator securely transfers PM data from the selected markets to Spectrum-NET.
3. Spectrum-NET ingests the data, analyzes and provides updates on anomalous interference events.
4. Spectrum Effect RF system analysts operate Spectrum-NET and provide 3 comprehensive reports to the operator.
5. Operator staff are provided with Spectrum-NET user accounts and access Spectrum-NET directly during the project.

Reference Materials
❖ Spectrum-NET RAN requirements
document and topology
spreadsheet.
❖ Spreadsheet of recommended PM
counters for the RAN vendor; raw
XML format is most desired.
❖ Spreadsheet with the Spectrum-NET
default KPI calculations.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 21
Spectrum-NET
Key Benefits: A Real Game Changer for Mobile Operators!
1. Automatically detects, characterizes, classifies, aggregates, locates, predicts and measures the impact of anomalous
interference.
2. Provides characteristics, location and classification for external interference events; invaluable for in-field hunting.
3. Automatically captures, characterizes and prioritizes PIM interference issues. Provides insights as to when PIM is
occurring and contributing frequencies. Enables effective testing, troubleshooting and repair of PIM interference issues
and ensures focusing on the most impactful issues.
4. Captures all incidence of cross border interference, including during periods of tropospheric ducting.
5. Includes a dashboard of the interference environment, breakdown of the interference measures from each source type
across each frequency band, the KPI impact from each interference event and analysis of KPI improvement with each
resolved interference event.
6. Prioritizes each anomalous interference event based on KPI impact.
7. Will prepare Operators for dealing with interference issues during their 5G NR rollouts.
• New 5G NR frequencies will increase the amount of PIM interference events. As operators add new bands for 5G NR, the potential for PIM
increases on operator’s existing UMTS and LTE channels. Likewise, the new 5G NR channels may also be impacted from PIM interference
contributed to by operator’s UMTS and LTE channels.
• Legacy devices (e.g., WiMAX) transmitting inside the 5G NR band allocations will generate additional external interference.
• 5G NR TDD channels will suffer from severe self-interference especially during periods of tropospheric ducting.
• This external, PIM and ducting uplink interference will substantially impact the uplink performance of 5G cells.
• Due to channel reciprocity applied for the downlink channel estimation, the downlink performance of 5G cells with uplink interference will also be
substantially impacted. This interference will need to be addressed in order to realize the expected benefits from new 5G NR technologies such
as Massive MIMO.

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 22
Spectrum-NET Audit Project
Key Deliverables
❖ Anomalous Interference Dashboard
❖ % of cells by RF band with anomalous interference
❖ Distribution of anomalous interference events: % external, RAN
generated, PIM, faulty hardware and TDD self-interference
❖ KPI impact of the anomalous interference events
❖ Comparison of markets with the global average
❖ Actionable Interference Events:
❖ Type: external, cross border, PIM, ducting, faulty hardware
and RAN generated interference
❖ Cells and sites impacted by the event
❖ Frequency characteristics
❖ % of time active during the project and temporal patterns
❖ Per-ROP activity %
❖ Locations of external interference sources
❖ Locations of cross border interference sources ❖ KPI Impact Assessment Metrics:
❖ For PIM interference, the contributing frequencies and ❖ LTE: ERAB Drop Rate, Service Setup Failure Rate, Handover
assistance with remote testing to confirm contributors Failure Rate, VoLTE Access Failure Rate and VoLTE Drop
Rate.
❖ TDD self-interference victim cells and aggressor cells
❖ UMTS: Accessibility CS, Accessibility PS, Drop Rate CS, Drop
❖ Measures of KPI impact Rate PS, Soft Handover Failure Rate

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 23
 Thank you!

10510 Northup Way Suite 130

Kirkland, WA 98033

www.spectrumeffect.com

© Copyright 2020 Spectrum Effect. All Rights Reserved. www.spectrumeffect.com Confidential & Proprietary 24