SlideShare une entreprise Scribd logo
1  sur  18
Télécharger pour lire hors ligne
CUSUM ANOMALY
DETECTION
1
THE CUSUM ANOMALY DETECTION ALGORITHM WAS CREATED IN
RESPONSE TO THE NEED FOR AN AUTOMATIZED METHOD OF
SEARCHING M-LAB’S VAST DATABASE OF NETWORK DIAGNOSTIC
TEST RESULTS
NOT FOR SINGLE OUTLIER POINTS, BUT FOR A SERIES OF
UNUSUALLY HIGH OR LOW MEASUREMENTS.
IT WAS DEVELOPED DURING THE COURSE OF A THREE MONTH
LONG “OUTREACHY” INTERNSHIP AT MEASUREMENT LAB IN THE
SUMMER OF 2015.
WHY WAS CAD CREATED?
2
3
M-LAB DATA EXAMPLES
4
260280300320340360
Vietnam −−Daily Median of Minimum Round Trip Time
Date
RoundTripTime
2013−09−01 2013−09−28 2013−10−25 2013−11−21 2013−12−18 2014−01−14 2014−02−10 2014−03−09 2014−04−05 2014−05−02
5
6
CUSUM CHARTS
• It is a statistical control chart
• It is a graph that is used to show how a process changes over
time.
• It has a center line for the average
• It has an upper control and a lower control line
• The CUSUM chart uses four parameters:
1. the expected mean of the process
2. the expected standard deviation of the process
3. the size of the shift that is to be detected — k
4. the control limit — H
CUSUM CHART EXAMPLE
7
8
APPLYING CUSUM CHARTS TO INTERNET
PERFORMANCE VARIABLES
Iran’s daily median throughput in 2013:
Iran’s daily median throughput in 2013
with the training set in red:
9
Properties of the training set:
10
The result of applying the CUSUM Chart to the Iran time series, using the mean and the standard
deviation of the training set as the expected mean and the training set:
11
CAD AND ITS DESIGN
OVERVIEW
The implementation of CAD was written in R and it uses the qcc* package to find the
CUSUM chart of a time series.
CAD uses the sliding window technique.
For each window, CAD searches the time series along the window
for a training set. If one is found, CAD applies the CUSUM chart to the entire time
series along the window.
After interpreting the results of the CUSUM chart, some of the points are designated
as possible anomalies.
This procedure is repeated for every window down the length of the time series.
The output of the process is the indices of the anomalies within the time series and a
graph of the time series with anomalies in red.
*Scrucca, L. (2004). qcc: an R package for quality control charting and statistical process control. R News 4/1, 11-17.
12
TUNING PARAMETERS
CAD is mostly automated, however, there are still a few parameters that, although
they have default values, can nonetheless be adjusted by the user. These are:
• lambda: the minimum length of the anomalous subsequences that CAD should
detect; its default value is 5.
• delta: an offset added to the CUSUM parameter K; its default value is 3.
• type: the choices for this parameter are upper or lower. It determines the type
of anomaly CAD should search for.
EXAMPLES AND RESULTS
13
The M-Lab Consortium Technical Report, ISP Interconnection and its Impact on
Consumer Internet Performance*, uncovered instances of performance degradation
in the US using M-Lab’s NDT datasets. CAD was tested on these time series, since
they contained known anomalies.
*http://www.measurementlab.net/static/observatory/M-Lab_Interconnection_Study_US.pdf
The focus of the examples is on the download throughput and packet retransmit
rate data from the New York City area, concerning the customers of Time Warner
Cable, Comcast, and Verizon connecting across the transit ISP Cogent.
These NDT time series span the time period from January 1, 2012 to September 30,
2014 . M-Lab’s report demonstrated the degradation of Internet performance
during the time period between April to June 2013 and late February 15 2014.
14
Settings:
type=lower
delta=3
lambda=5
Settings:
type=lower
delta=3
lambda=5
15
Settings:
type=lower
delta=3
lambda=5
Settings:
type=lower
delta=5
lambda=5
16
Settings:
type=upper
delta=3
lambda=1
Settings:
type=upper
delta=3
lambda=3
17
Settings:
type=upper
delta=5
lambda=5
The average false negative error rate for these examples was about 0.036.
The average false positive error rate for these examples was about 0.008.
EVALUATION:
CONCLUSION & FUTURE WORK
• The CAD method works well for discovering anomalies in network performance
data, with a high rate of successful anomaly detection and a low rate of
false positives
• CAD has not been tested outside a narrow scope
• CAD is an automatic process, but it does not provide a list of points that can
be labeled anomalous with absolute certainty. It merely provides a potential list of
anomalies that the user must assess and then adjust parameters as necessary.
• Future work will be focused on eliminating delta from the list of tunable
parameters and finding a way to make this method into a map-reduce process
18

Contenu connexe

Tendances

Searching for Dynamical Resemblance Between Time Series: A Method Based on No...
Searching for Dynamical Resemblance Between Time Series: A Method Based on No...Searching for Dynamical Resemblance Between Time Series: A Method Based on No...
Searching for Dynamical Resemblance Between Time Series: A Method Based on No...
Gladstone Alves
 
NEAL-2016 ARL Symposium Poster
NEAL-2016 ARL Symposium PosterNEAL-2016 ARL Symposium Poster
NEAL-2016 ARL Symposium Poster
Barbara Jean Neal
 
Multisensor data fusion in object tracking applications
Multisensor data fusion in object tracking applicationsMultisensor data fusion in object tracking applications
Multisensor data fusion in object tracking applications
Sayed Abulhasan Quadri
 
Sales forecasting of an airline company using time series analysis (1) (1)
Sales forecasting of an airline company using time series analysis (1) (1)Sales forecasting of an airline company using time series analysis (1) (1)
Sales forecasting of an airline company using time series analysis (1) (1)
Ashish Ranjan
 
Investigating WetSpa model application in the Distributed Model Intercomparis...
Investigating WetSpa model application in the Distributed Model Intercomparis...Investigating WetSpa model application in the Distributed Model Intercomparis...
Investigating WetSpa model application in the Distributed Model Intercomparis...
Alireza Safari
 

Tendances (20)

Time_Series_Assignment
Time_Series_AssignmentTime_Series_Assignment
Time_Series_Assignment
 
ROUTING PRESSURE: A CHANNEL-RELATED AND TRAFFIC-AWARE METRIC OF ROUTING ALGOR...
ROUTING PRESSURE: A CHANNEL-RELATED AND TRAFFIC-AWARE METRIC OF ROUTING ALGOR...ROUTING PRESSURE: A CHANNEL-RELATED AND TRAFFIC-AWARE METRIC OF ROUTING ALGOR...
ROUTING PRESSURE: A CHANNEL-RELATED AND TRAFFIC-AWARE METRIC OF ROUTING ALGOR...
 
Searching for Dynamical Resemblance Between Time Series: A Method Based on No...
Searching for Dynamical Resemblance Between Time Series: A Method Based on No...Searching for Dynamical Resemblance Between Time Series: A Method Based on No...
Searching for Dynamical Resemblance Between Time Series: A Method Based on No...
 
Fall detection
Fall detectionFall detection
Fall detection
 
NEAL-2016 ARL Symposium Poster
NEAL-2016 ARL Symposium PosterNEAL-2016 ARL Symposium Poster
NEAL-2016 ARL Symposium Poster
 
Lecture 09: SLAM
Lecture 09: SLAMLecture 09: SLAM
Lecture 09: SLAM
 
Interpolation 2013
Interpolation 2013Interpolation 2013
Interpolation 2013
 
3D Analyst - Watershed
3D Analyst - Watershed3D Analyst - Watershed
3D Analyst - Watershed
 
Multisensor data fusion in object tracking applications
Multisensor data fusion in object tracking applicationsMultisensor data fusion in object tracking applications
Multisensor data fusion in object tracking applications
 
Change Point Analysis
Change Point AnalysisChange Point Analysis
Change Point Analysis
 
XL-Miner: Timeseries
XL-Miner: TimeseriesXL-Miner: Timeseries
XL-Miner: Timeseries
 
Template attack versus Bayes classifier
Template attack  versus Bayes classifierTemplate attack  versus Bayes classifier
Template attack versus Bayes classifier
 
3D Analyst - Watershed and Stream Network
3D Analyst - Watershed and Stream Network3D Analyst - Watershed and Stream Network
3D Analyst - Watershed and Stream Network
 
Sales Data Forecasting for Airline
Sales Data Forecasting for AirlineSales Data Forecasting for Airline
Sales Data Forecasting for Airline
 
New three dimensional space vector based switching signal generation techniqu...
New three dimensional space vector based switching signal generation techniqu...New three dimensional space vector based switching signal generation techniqu...
New three dimensional space vector based switching signal generation techniqu...
 
Sales forecasting of an airline company using time series analysis (1) (1)
Sales forecasting of an airline company using time series analysis (1) (1)Sales forecasting of an airline company using time series analysis (1) (1)
Sales forecasting of an airline company using time series analysis (1) (1)
 
Shantanu Kulkarni Research Career Interests
Shantanu Kulkarni Research Career InterestsShantanu Kulkarni Research Career Interests
Shantanu Kulkarni Research Career Interests
 
Brussels airport forecast
Brussels airport  forecast Brussels airport  forecast
Brussels airport forecast
 
FINAL [Autosaved]
FINAL [Autosaved]FINAL [Autosaved]
FINAL [Autosaved]
 
Investigating WetSpa model application in the Distributed Model Intercomparis...
Investigating WetSpa model application in the Distributed Model Intercomparis...Investigating WetSpa model application in the Distributed Model Intercomparis...
Investigating WetSpa model application in the Distributed Model Intercomparis...
 

En vedette

Lecture 14 cusum and ewma
Lecture 14 cusum and ewmaLecture 14 cusum and ewma
Lecture 14 cusum and ewma
Ingrid McKenzie
 
Anomaly Detection Via PCA
Anomaly Detection Via PCAAnomaly Detection Via PCA
Anomaly Detection Via PCA
Deepak Kumar
 
Solar collectors nces
Solar collectors ncesSolar collectors nces
Solar collectors nces
Anu71
 
Solar energy ppt
Solar energy pptSolar energy ppt
Solar energy ppt
shubhajit_b
 

En vedette (15)

Lecture 14 cusum and ewma
Lecture 14 cusum and ewmaLecture 14 cusum and ewma
Lecture 14 cusum and ewma
 
Isen 614 project report
Isen 614 project reportIsen 614 project report
Isen 614 project report
 
SPC and Control Charts
SPC and Control ChartsSPC and Control Charts
SPC and Control Charts
 
Isen 614 project presentation
Isen 614 project presentationIsen 614 project presentation
Isen 614 project presentation
 
Solar Energy Collectors
Solar Energy CollectorsSolar Energy Collectors
Solar Energy Collectors
 
Statistical Process Control
Statistical Process ControlStatistical Process Control
Statistical Process Control
 
Anomaly Detection Via PCA
Anomaly Detection Via PCAAnomaly Detection Via PCA
Anomaly Detection Via PCA
 
An Introduction into Anomaly Detection Using CUSUM
An Introduction into Anomaly Detection Using CUSUMAn Introduction into Anomaly Detection Using CUSUM
An Introduction into Anomaly Detection Using CUSUM
 
Variable control chart
Variable control chartVariable control chart
Variable control chart
 
Solar collectors nces
Solar collectors ncesSolar collectors nces
Solar collectors nces
 
Solar flat plate collector
Solar flat plate collectorSolar flat plate collector
Solar flat plate collector
 
Statistical Process Control
Statistical Process ControlStatistical Process Control
Statistical Process Control
 
Statistical Quality Control.
Statistical Quality Control.Statistical Quality Control.
Statistical Quality Control.
 
LIÇÃO 12 - QUEM AMA CUMPRE PLENAMENTE A LEI DIVINA
LIÇÃO 12 - QUEM AMA CUMPRE PLENAMENTE A LEI DIVINALIÇÃO 12 - QUEM AMA CUMPRE PLENAMENTE A LEI DIVINA
LIÇÃO 12 - QUEM AMA CUMPRE PLENAMENTE A LEI DIVINA
 
Solar energy ppt
Solar energy pptSolar energy ppt
Solar energy ppt
 

Similaire à CAD -- CUSUM Anomaly Detection

Congestion control in computer networks using a modified red aqm algorithm
Congestion control in computer networks using a modified red aqm algorithmCongestion control in computer networks using a modified red aqm algorithm
Congestion control in computer networks using a modified red aqm algorithm
eSAT Journals
 
Online flooding monitoring in packed towers
Online flooding monitoring in packed towersOnline flooding monitoring in packed towers
Online flooding monitoring in packed towers
James Cao
 
Poster_Reseau_Neurones_Journees_2013
Poster_Reseau_Neurones_Journees_2013Poster_Reseau_Neurones_Journees_2013
Poster_Reseau_Neurones_Journees_2013
Pedro Lopes
 

Similaire à CAD -- CUSUM Anomaly Detection (20)

Congestion control in computer networks using a modified red aqm algorithm
Congestion control in computer networks using a modified red aqm algorithmCongestion control in computer networks using a modified red aqm algorithm
Congestion control in computer networks using a modified red aqm algorithm
 
Congestion control in computer networks using a
Congestion control in computer networks using aCongestion control in computer networks using a
Congestion control in computer networks using a
 
Congestion control in computer networks using a
Congestion control in computer networks using aCongestion control in computer networks using a
Congestion control in computer networks using a
 
Jgrass newage-water
Jgrass newage-waterJgrass newage-water
Jgrass newage-water
 
Improving continuous process operation using data analytics delta v applicati...
Improving continuous process operation using data analytics delta v applicati...Improving continuous process operation using data analytics delta v applicati...
Improving continuous process operation using data analytics delta v applicati...
 
ESTIMATING HANDLING TIME OF SOFTWARE DEFECTS
ESTIMATING HANDLING TIME OF SOFTWARE DEFECTSESTIMATING HANDLING TIME OF SOFTWARE DEFECTS
ESTIMATING HANDLING TIME OF SOFTWARE DEFECTS
 
Online flooding monitoring in packed towers
Online flooding monitoring in packed towersOnline flooding monitoring in packed towers
Online flooding monitoring in packed towers
 
Smart E-Logistics for SCM Spend Analysis
Smart E-Logistics for SCM Spend AnalysisSmart E-Logistics for SCM Spend Analysis
Smart E-Logistics for SCM Spend Analysis
 
Poster_Reseau_Neurones_Journees_2013
Poster_Reseau_Neurones_Journees_2013Poster_Reseau_Neurones_Journees_2013
Poster_Reseau_Neurones_Journees_2013
 
SD-GCM Formulas
SD-GCM FormulasSD-GCM Formulas
SD-GCM Formulas
 
proposedfaultdetectiononoverheadtransmissionlineusingparticleswarmoptimizatio...
proposedfaultdetectiononoverheadtransmissionlineusingparticleswarmoptimizatio...proposedfaultdetectiononoverheadtransmissionlineusingparticleswarmoptimizatio...
proposedfaultdetectiononoverheadtransmissionlineusingparticleswarmoptimizatio...
 
PROPOSED FAULT DETECTION ON OVERHEAD TRANSMISSION LINE USING PARTICLE SWARM ...
PROPOSED FAULT DETECTION ON OVERHEAD TRANSMISSION LINE  USING PARTICLE SWARM ...PROPOSED FAULT DETECTION ON OVERHEAD TRANSMISSION LINE  USING PARTICLE SWARM ...
PROPOSED FAULT DETECTION ON OVERHEAD TRANSMISSION LINE USING PARTICLE SWARM ...
 
Empirical Analysis of Radix Sort using Curve Fitting Technique in Personal Co...
Empirical Analysis of Radix Sort using Curve Fitting Technique in Personal Co...Empirical Analysis of Radix Sort using Curve Fitting Technique in Personal Co...
Empirical Analysis of Radix Sort using Curve Fitting Technique in Personal Co...
 
7-White Box Testing.ppt
7-White Box Testing.ppt7-White Box Testing.ppt
7-White Box Testing.ppt
 
A0250109
A0250109A0250109
A0250109
 
IRJET- Smart Railway System using Trip Chaining Method
IRJET- Smart Railway System using Trip Chaining MethodIRJET- Smart Railway System using Trip Chaining Method
IRJET- Smart Railway System using Trip Chaining Method
 
STATISTICAL PROCESS CONTROL(PPT).pptx
STATISTICAL PROCESS CONTROL(PPT).pptxSTATISTICAL PROCESS CONTROL(PPT).pptx
STATISTICAL PROCESS CONTROL(PPT).pptx
 
Applications of machine learning in Wireless sensor networks.
Applications of machine learning in Wireless sensor networks.Applications of machine learning in Wireless sensor networks.
Applications of machine learning in Wireless sensor networks.
 
IRJET- Different Data Mining Techniques for Weather Prediction
IRJET-  	  Different Data Mining Techniques for Weather PredictionIRJET-  	  Different Data Mining Techniques for Weather Prediction
IRJET- Different Data Mining Techniques for Weather Prediction
 
A REVIEW ON OPTIMIZATION OF LEAST SQUARES SUPPORT VECTOR MACHINE FOR TIME SER...
A REVIEW ON OPTIMIZATION OF LEAST SQUARES SUPPORT VECTOR MACHINE FOR TIME SER...A REVIEW ON OPTIMIZATION OF LEAST SQUARES SUPPORT VECTOR MACHINE FOR TIME SER...
A REVIEW ON OPTIMIZATION OF LEAST SQUARES SUPPORT VECTOR MACHINE FOR TIME SER...
 

Dernier

Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
amitlee9823
 
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
amitlee9823
 
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
amitlee9823
 
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
amitlee9823
 
FESE Capital Markets Fact Sheet 2024 Q1.pdf
FESE Capital Markets Fact Sheet 2024 Q1.pdfFESE Capital Markets Fact Sheet 2024 Q1.pdf
FESE Capital Markets Fact Sheet 2024 Q1.pdf
MarinCaroMartnezBerg
 
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get CytotecAbortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Riyadh +966572737505 get cytotec
 
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
amitlee9823
 
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
amitlee9823
 
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service BangaloreCall Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
amitlee9823
 

Dernier (20)

Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
Digital Advertising Lecture for Advanced Digital & Social Media Strategy at U...
 
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
Call Girls Bannerghatta Road Just Call 👗 7737669865 👗 Top Class Call Girl Ser...
 
Accredited-Transport-Cooperatives-Jan-2021-Web.pdf
Accredited-Transport-Cooperatives-Jan-2021-Web.pdfAccredited-Transport-Cooperatives-Jan-2021-Web.pdf
Accredited-Transport-Cooperatives-Jan-2021-Web.pdf
 
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
Call Girls Hsr Layout Just Call 👗 7737669865 👗 Top Class Call Girl Service Ba...
 
Call Girls in Sarai Kale Khan Delhi 💯 Call Us 🔝9205541914 🔝( Delhi) Escorts S...
Call Girls in Sarai Kale Khan Delhi 💯 Call Us 🔝9205541914 🔝( Delhi) Escorts S...Call Girls in Sarai Kale Khan Delhi 💯 Call Us 🔝9205541914 🔝( Delhi) Escorts S...
Call Girls in Sarai Kale Khan Delhi 💯 Call Us 🔝9205541914 🔝( Delhi) Escorts S...
 
Anomaly detection and data imputation within time series
Anomaly detection and data imputation within time seriesAnomaly detection and data imputation within time series
Anomaly detection and data imputation within time series
 
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
Call Girls Jalahalli Just Call 👗 7737669865 👗 Top Class Call Girl Service Ban...
 
Generative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and MilvusGenerative AI on Enterprise Cloud with NiFi and Milvus
Generative AI on Enterprise Cloud with NiFi and Milvus
 
Call me @ 9892124323 Cheap Rate Call Girls in Vashi with Real Photo 100% Secure
Call me @ 9892124323  Cheap Rate Call Girls in Vashi with Real Photo 100% SecureCall me @ 9892124323  Cheap Rate Call Girls in Vashi with Real Photo 100% Secure
Call me @ 9892124323 Cheap Rate Call Girls in Vashi with Real Photo 100% Secure
 
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
Mg Road Call Girls Service: 🍓 7737669865 🍓 High Profile Model Escorts | Banga...
 
BabyOno dropshipping via API with DroFx.pptx
BabyOno dropshipping via API with DroFx.pptxBabyOno dropshipping via API with DroFx.pptx
BabyOno dropshipping via API with DroFx.pptx
 
FESE Capital Markets Fact Sheet 2024 Q1.pdf
FESE Capital Markets Fact Sheet 2024 Q1.pdfFESE Capital Markets Fact Sheet 2024 Q1.pdf
FESE Capital Markets Fact Sheet 2024 Q1.pdf
 
Midocean dropshipping via API with DroFx
Midocean dropshipping via API with DroFxMidocean dropshipping via API with DroFx
Midocean dropshipping via API with DroFx
 
BigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptxBigBuy dropshipping via API with DroFx.pptx
BigBuy dropshipping via API with DroFx.pptx
 
(NEHA) Call Girls Katra Call Now 8617697112 Katra Escorts 24x7
(NEHA) Call Girls Katra Call Now 8617697112 Katra Escorts 24x7(NEHA) Call Girls Katra Call Now 8617697112 Katra Escorts 24x7
(NEHA) Call Girls Katra Call Now 8617697112 Katra Escorts 24x7
 
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get CytotecAbortion pills in Doha Qatar (+966572737505 ! Get Cytotec
Abortion pills in Doha Qatar (+966572737505 ! Get Cytotec
 
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
Junnasandra Call Girls: 🍓 7737669865 🍓 High Profile Model Escorts | Bangalore...
 
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
Call Girls Indiranagar Just Call 👗 7737669865 👗 Top Class Call Girl Service B...
 
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service BangaloreCall Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
Call Girls Begur Just Call 👗 7737669865 👗 Top Class Call Girl Service Bangalore
 
Mature dropshipping via API with DroFx.pptx
Mature dropshipping via API with DroFx.pptxMature dropshipping via API with DroFx.pptx
Mature dropshipping via API with DroFx.pptx
 

CAD -- CUSUM Anomaly Detection

  • 2. THE CUSUM ANOMALY DETECTION ALGORITHM WAS CREATED IN RESPONSE TO THE NEED FOR AN AUTOMATIZED METHOD OF SEARCHING M-LAB’S VAST DATABASE OF NETWORK DIAGNOSTIC TEST RESULTS NOT FOR SINGLE OUTLIER POINTS, BUT FOR A SERIES OF UNUSUALLY HIGH OR LOW MEASUREMENTS. IT WAS DEVELOPED DURING THE COURSE OF A THREE MONTH LONG “OUTREACHY” INTERNSHIP AT MEASUREMENT LAB IN THE SUMMER OF 2015. WHY WAS CAD CREATED? 2
  • 3. 3
  • 5. 260280300320340360 Vietnam −−Daily Median of Minimum Round Trip Time Date RoundTripTime 2013−09−01 2013−09−28 2013−10−25 2013−11−21 2013−12−18 2014−01−14 2014−02−10 2014−03−09 2014−04−05 2014−05−02 5
  • 6. 6 CUSUM CHARTS • It is a statistical control chart • It is a graph that is used to show how a process changes over time. • It has a center line for the average • It has an upper control and a lower control line • The CUSUM chart uses four parameters: 1. the expected mean of the process 2. the expected standard deviation of the process 3. the size of the shift that is to be detected — k 4. the control limit — H
  • 8. 8 APPLYING CUSUM CHARTS TO INTERNET PERFORMANCE VARIABLES Iran’s daily median throughput in 2013: Iran’s daily median throughput in 2013 with the training set in red:
  • 9. 9 Properties of the training set:
  • 10. 10 The result of applying the CUSUM Chart to the Iran time series, using the mean and the standard deviation of the training set as the expected mean and the training set:
  • 11. 11 CAD AND ITS DESIGN OVERVIEW The implementation of CAD was written in R and it uses the qcc* package to find the CUSUM chart of a time series. CAD uses the sliding window technique. For each window, CAD searches the time series along the window for a training set. If one is found, CAD applies the CUSUM chart to the entire time series along the window. After interpreting the results of the CUSUM chart, some of the points are designated as possible anomalies. This procedure is repeated for every window down the length of the time series. The output of the process is the indices of the anomalies within the time series and a graph of the time series with anomalies in red. *Scrucca, L. (2004). qcc: an R package for quality control charting and statistical process control. R News 4/1, 11-17.
  • 12. 12 TUNING PARAMETERS CAD is mostly automated, however, there are still a few parameters that, although they have default values, can nonetheless be adjusted by the user. These are: • lambda: the minimum length of the anomalous subsequences that CAD should detect; its default value is 5. • delta: an offset added to the CUSUM parameter K; its default value is 3. • type: the choices for this parameter are upper or lower. It determines the type of anomaly CAD should search for.
  • 13. EXAMPLES AND RESULTS 13 The M-Lab Consortium Technical Report, ISP Interconnection and its Impact on Consumer Internet Performance*, uncovered instances of performance degradation in the US using M-Lab’s NDT datasets. CAD was tested on these time series, since they contained known anomalies. *http://www.measurementlab.net/static/observatory/M-Lab_Interconnection_Study_US.pdf The focus of the examples is on the download throughput and packet retransmit rate data from the New York City area, concerning the customers of Time Warner Cable, Comcast, and Verizon connecting across the transit ISP Cogent. These NDT time series span the time period from January 1, 2012 to September 30, 2014 . M-Lab’s report demonstrated the degradation of Internet performance during the time period between April to June 2013 and late February 15 2014.
  • 17. 17 Settings: type=upper delta=5 lambda=5 The average false negative error rate for these examples was about 0.036. The average false positive error rate for these examples was about 0.008. EVALUATION:
  • 18. CONCLUSION & FUTURE WORK • The CAD method works well for discovering anomalies in network performance data, with a high rate of successful anomaly detection and a low rate of false positives • CAD has not been tested outside a narrow scope • CAD is an automatic process, but it does not provide a list of points that can be labeled anomalous with absolute certainty. It merely provides a potential list of anomalies that the user must assess and then adjust parameters as necessary. • Future work will be focused on eliminating delta from the list of tunable parameters and finding a way to make this method into a map-reduce process 18