Download Cognitive Radio Networking in the ISM Band...
Cognitive Radio Networking in the ISM Band John Sydor Siva Palaninathan Bernard Doray Jianxing Hu The Research Broad Band Wireless Group
www.crc.ca/coral
© 2011 In the Name of the Her Majesty the Queen of Canada
Webinar Overview
Overview of Cognitive Radio concepts. The CRC-CORAL Wi-Fi Cognitive Radio Network Platform. Implementation of Cognitive control Software for control of sensing and WiFi packet emission in space, time, & channel The Radio Environment Awareness Map (REAM): Use of Cognitive Engines to control the Network. Applications: Dynamic Spectrum Access & Data mining in the REAM
©
Generic Wireless Network Wireless Network Data Sources
Radio Layer Emission Control EIRP, direction, timing, modulation rate, channel, Radio Environment Sensors, MAC layer protocol variables
Uncooperative Foreign Interferers Primary User
Channel Medium Interference: intensity, location, propagation User Population: distribution & access behavior
Radio Layer Reception Control & Performance Radio Environment Sensors
©
Wireless Network Data Sinks
Cooperative Interferers
Formative Cognitive Radio Architecture Performance figure of merit (Bits/Hz/M2); stability criterion, bandwidth allocation fairness,etc
Wireless Network Data Sources
Emission & location Information Radio Resource Requirements
Radio Layer Emission Control EIRP, direction, timing, modulation rate, channel, Radio Environment Sensors, MAC layer protocol variables
Cognitive Engine Statistical Analyzer Adaptive Intelligence Data Mining
Control
Policy or etiquette, data base inquiry requirement.
Historical Sensor Data
Radio Environment Sensed Information
Uncooperative Foreign Interferers Primary User
Channel Medium Interference: intensity, location, propagation User Population: distribution & access behavior
Radio Layer Reception Control & Performance Radio Environment Sensors
©
Wireless Network Data Sinks
Radio Environment Map Cooperative Interferers
Collaborative Control
Radio Environment Sensed Information Control
Partitioning the CR Wireless Network Data Sources
Performance figure of merit (Bits/Hz/M2); stability criterion, bandwidth allocation fairness,etc
Policy or etiquette, data base inquiry requirement. Radio Layer Emission Control EIRP, direction, timing, modulation rate, channel, Radio Environment Sensors, MAC layer protocol variables
Channel Medium Interference: intensity, location, propagation User Population: distribution & access behavior
Cognitive Engine Statistical Analyzer Adaptive Intelligence Data Mining
Radio Environment Map
CR_NMS Domain Radio Layer Reception Control & Performance Radio Environment Sensors
©
Wireless Network Data Sinks
WiFi_CR Terminal Domain
Partitioned Components of CORAL
WiFi_CR Terminal (Phy/802.11 functions)
TCP/IP CR_NMS (Cognitive Radio Engines)
©
Examples of Modular Deployments
Single CRN on one CR_NMS ©
Two CRNs on one CR_NMS
Examples of Modular Deployments
4 CRNs on 4 CR_NMS ©
The CRC-CORAL Wi-Fi Cognitive Radio Network Platform WiFi_CR Layout
19 pin connector to Break Out Board
GPS Antenna
CRC Router Board Radio Cards Spectrum Board
Environmentally Sealed case
2 monopole 2.4 GHz antennas
Ethernet Connector to POE Unit or existing power connection
CRC EB Board Wall/Pole Mounting
©
PoE Unit
The CRC-CORAL Wi-Fi Cognitive Radio Network Platform PHY Layer Emission Control
PHY Layer Capabilities of WiFi_CR Standard IEEE 802.11g operation, Infrastructure Mode TDD/TDMA constrained CSMA/CA Per packet beam steering (8 Beams) Per Slot antenna beam steering Control of EIRP, channel, Ack policy, modulation rate,etc GPS, Beacon based, and ARP based TDD slot synchronization
©
The CRC-CORAL Wi-Fi Cognitive Radio Network Platform WiFi_CR: block signal processing subsystem layout GPS Antenna Antenna Steering/Switch Control. RF/Tx TDM Control & TX power detection.
802.11g Beacon announcement
Omni Antennas
RS232
WiFi sensor radio card. IEEE 802.3 Ethernet Ethernet Buffer Board (EBB).
IEEE 802.3 Ethernet
Antenna Switch Option
RF Diversity Rx
RF Diversity Port Rx
Madwifi 4 port wireless Router mother board.
Diversity 8 antenna 8 port RF array switch.
RF Link Interface circuits. RF Port Tx/Rx WiFi data link radio card.
8 port RF switch.
8 antenna
array
RF Tx/Rx
60 GHz RF & steerable array antenna.
RF/Tx TDM Control & Framing signals for control of co-located WIFI_CR AP or Client devices RS232
2400-2500 MHz spectrum analyzer. RF Diversity Rx
Deployment specific RF and detection system modules.
5.5-5.8 GHz RF & steerable array antenna.
Primary User Matched Filter detector.
Dotted lines for future enhancements
©
Primary user Cyclostationary signal detector
600-700 MHz RF & steerable array antenna.
The CRC-CORAL Wi-Fi Cognitive Radio Network Platform TDD/TDMA synchronized slots
TDD Slots
AP
CL
WiFi Packets
WiFi TDD with No ACK data transmission between An Access Point and Client terminal.
©
Multi-Sector Beam Antenna
6 directional Sector/1 omni-sector Array with diversity.
8X1 RF Switch
©
Example:Relay and Multiple Beam Steering Configurations
©
Agenda - RouterBoard
RouterBoard Hardware RouterBoard Software Wireless Sniffer
©
RouterBoard Hardware
Mikrotik RB433 (AH, UAH)
©
MIPS-based AR7100 300 MHz 64 MB RAM 64 MB NAND 3 Ethernet ports, 1 PoE, Auto MDI/X 1 Serial port LED GPIO, used for beacon alerts 3 MiniPCI slots
RouterBoard Hardware cont.
Wistron CM9 MiniPCI 80211a/b/g Atheros AR5212 based Infrastructure mode – Link Interface Monitor mode – Sniffer Interface Madwifi Driver
©
RouterBoard Software
©
•
RouterBOOT Booter
•
Linux – Customized OpenWRT
•
Modified Madwifi Driver
•
NetSNMP
•
Wireless sniffer
(Kismet Based)
RouterBOOT Booter Available via serial port by pressing DELETE key during boot cycle
BOOT Device Menu your choice: o - boot device Select boot device: e - boot over Ethernet * n - boot from NAND, if fail then Ethernet 1 - boot Ethernet once, then NAND o - boot from NAND only b - boot chosen device
©
Board Info Menu Board type: 433 Serial number: 21FE01987F32 Firmware version: 2.20 CPU frequency: 300 MHz Memory size: 64 MB eth1 MAC address: 00:0C:42:45:27:85 eth2 MAC address: 00:0C:42:45:27:86 eth3 MAC address: 00:0C:42:45:27:87
Linux - OpenWRT
• • •
Kernel 2.6.32 Kamikaze 8_09 Real time extension Modifications for CORAL
• •
©
GPIO for Beacon alerts Various init scripts
Madwifi Driver
Core version r3314 Latest patches by the development team Patches related to cross compilation
Modifications for CORAL
©
Beacon notification Best effort Q for all type of traffic Number of retries Diversity enabled Deterministic transmission; CCA, backoff CTS/RTS Disabling Turn ON/OFF ACK policy
NetSNMP
©
NetSNMP cont.
Core version 5.1.2 Version 2 Latest patches by the development team Patches related to cross compilation Modifications for CORAL
©
802.11 MIB agent implemented/modified • Handling all radio related configuration commands • Link related statistics collection A new sniffer agent implemented • Wireless sniffer • Handling part of EBB communication • Various CORAL system specific commands
Wireless Sniffer
• • • • • • •
©
Kismet based, significantly modified for on-demand scanning Supports Canada and Europe bands Madwifi in monitor mode Captures raw 802.11 packets only Corrupted packets are accounted Controlled by SNMP Highly customizable
Wireless Sniffer – Interference capturing process 500 ms per channel Interference bins are created for each unique set of:
• •
When a packet is received, its bin is updated with:
•
©
Source MAC Destination MAC BSSID Channel Packet Type Packet Subtype
timestamp number of packets accumulated in the bin average RSSI packet utilization (packet length/packet transmitted rate)
Wireless Sniffer – Interference capturing process cont. •
Information captured from a packet
©
Captured time; local to system clock GPS Location of the sniffer node Packet type/subtype Sniffer node ID TX Duty cycle TX Avg Channel the packet is detected on Channel the sniffer is on Source MAC address Destination MAC address Total duration of packet(s) SSID RSSI or averaged RSSI Number of packets Transmit rate of the packet
Wireless Sniffer – customization •
Sniffer software can be customized for:
©
Packets only from a specific node Management, controls or data packets only Complete decoded-preamble Channel utilization Statistics collection: Number of corrupted packets, Retransmission packets, etc
Wireless Sniffer cont.
©
High CPU usage All packets reaching monitor interface are accounted Fields extraction Utilization calculation RAM usage
CRNMS
©
CRNMS API Interface provided by the CRNMS to control / interrogate CORAL terminals Interface provided to access the REAM data collected by the CRNMS API Available for the following programming languages: C, MATLAB, Python APIs specified/generated from the WSDL (Web Service Description Language) specification
©
CRNMS API Interface used for scheduling, monitoring, configuration,… setting direction, data rate, EIRP, channel, scheduling soliciting interference, position, occupancy
©
CRNMS API getVersionSynch getNodeInfoSynch getGpsInfo getStatusInfo setPollingInterval collectData resetEbb ©
setBestChannel setCodeWord setTxPower setLinkSpeed setAntennaDstTable setAck setEbbMode
MATLAB Toolbox
©
Examples of how the API can be used filter data in REAM to store specific network characteristics data mining interference signature versus time study of network activity tune network to maximize throughput dynamically / autonomously sense interference, adjust timeslots, beam angles, channel to avoid interference (spatial, temporal, spectrum tuning) find whitespace and use it opportunistically
©
REAM Database • Interference Table • WIFI interference collected by the CORAL terminals (SA, DA, SSID, RSSI, Packet type, Packet Subtype, ...) • Spectrum Data Table Spectrum Analyzer data collected by the CORAL terminals (101 measurements from 2.4 to 2.5Ghz) • Nodes Table • Information about the nodes currently part of the Cognitive Radio Network (including location if available) • Alerts Table • Log of primary users detection ©
SQL REAM database can be queried using the SQL language; examples:
©
SELECT * FROM interference WHERE channel = 11 ; SELECT * FROM interference WHERE channel = 11 and rssi > -65 ; SELECT key,bssid,sa,da,ssid,rssi FROM interference where mode = 'AP'; SELECT DISTINCT sa FROM interference WHERE rssi > -65 ; SELECT DISTINCT node from nodes WHERE parent = '12:34:56:78:90:02' ORDER by node; SELECT sa,da,rssi FROM interference WHERE rssi > -60 AND sa IN (SELECT DISTINCT node from nodes WHERE parent = '12:34:56:78:90:00' ) ORDER by sa ; SELECT * from interference WHERE time > ( now() - INTERVAL '1 minute' ) ;
GUI – Spectrum Analyzer Data
©
GUI – Mapping Capabilities
©
Application Example: Investigation of the Outdoor Urban Interference environment by mining The REAM data base….urban target area for the experiment
Google-View of the Test Site In Ottawa, WiFi_CR units in Yellow
©
Application Example: Investigation of the Outdoor Urban Interference environment by mining the REAM data base…experiment set up.
WiFi_CR Terminals deployed, all in LOS of each other
©
Application Example: Investigation of the Outdoor Urban Interference environment by mining the REAM data base: Extracted Results…Occupancy by interference
Spectrum ‘Holes’
Interferers detected per 10 minute interval With packet powers>-62 dBm, for all sniffers
©
Packet Occupancy, all interfering packets With power >-82 dBm, as seen at each sniffer
Application Example: Investigation of the Outdoor Urban Interference environment by mining the REAM data base: Extracted Results…Occupancy variations
Total Occupancy and Occupancy by different types of WiFi Packets as seen on Sniffer 5, for powers > -82 dBm
©
Application Example: Investigation of the Outdoor Urban Interference environment by mining the REAM data base: Extracted Results…degrees of occupancy by users
User activity per sniffer over a 10 minutes sample. Users are unique source addresses; ~ same time, But at different sniffers
©
CDF of user activity over a 10 minutes sample. At different sniffers, power >-82 dBm
Application Example: Investigation of the Outdoor Urban Interference environment by mining the REAM data base: Extracted Results…correlation of interferenc
Top Graph: RSSI of received beacons emanating from terminals 1-6,8 as measured on terminal 7 Bottom Graph: Correlation between beacons’ RSSI emanating from terminal 6 and Terminals (1-7,8) as measured on Terminal 7
©
Top Graph: Correlation of beacons emanating from Terminal 1 as received at Terminals pairs {7,7},{7,6},{7,2},{7,3} Bottom Graph: Correlation of beacons emanating from Terminal 4 as received at Terminals {7,7},{7,6},{7,2},{7,3} 60 Minute duration, distances between terminal pairs shown
Dynamic Spectrum Access
Objective: Implement an algorithm that chooses the WiFi channel providing the best throughput in a long range point to point WiFi link. Approach: This DSA algorithm is to be based on the per channel interference energy and occupancy.
3 Km Interference from foreign WiFi NLOS, vegetation blocked path
CRC with CR_NMS
©
CSC Building with Client Radio
Dynamic Spectrum Access DSA Process
1.
2.
3.
4.
5.
©
CR NMS collects WiFi interference data from Coral terminals periodically Collected data are stored in database (REAM database) Channel selection application selects a best channel periodically according to the WiFi interference environment Application sends switch channel command to CR NMS when a better channel is selected other than current channel CR NMS sends command to AP to switch Channel
WiFi Interference
(1)
(5) (2) REAM CR_NM S
Coral API (3) (4)
Channel Selection Cognitive Engine
Dynamic Spectrum Access
Algorithm N
*
Selected Channel :
J = argmax { ISj(Ap) / 2 + j
The raw data representing the interference environment of CRN.
Channel
Node
RSSI (dbm)
Duration (us)
….
12:00
1
Ap
-70
2220
….
12:00
2
Sta1
-81
4002
….
12:01
2
Sta2
-75
12604
….
….
….
….
….
….
….
Uj =
(
Σ
(Dj,t) )
j =1, t = DTDI
/
Step 1
( STR * DI )
Energy for channel j is calculated as total energy that the WiFi interference packets have. M, DT
Ej =
Σ
(Dj,t * 10 ( RSSI / 10))
j =1, t = DT-DI
©
j,t
j
Channel (j)
Utilization (Uj) (%)
Energy (Ej)
1
20
3.07
2
9
1.2
3
10
0.4
….
….
….
Interference Score (IS) for Channel j IS j =
Channel (j)
Utilization (Uj) (%)
Energy (Ej)
1
10
2.028
2
6
0.34
3
3
0.205
….
….
….
Sta2 12:00~12:05 Channel (j)
Utilization (Uj) (%)
Energy (Ej)
1
30
0.928
2
8
0. 49
3
11
1.705
….
….
….
…
f
(U j ) –
g (E
j
)
12:00~12:05 Interference Score Table
Sta1 12:00~12:05
Utilization for channel j is calculated as percentage of the total duration of WiFi interference packets over the total scanned time. M, DT
k=1
Ap 12:00~12:05
CRNMS Interference DB Time
Σ ( IS (Stak) / (2*N) }
Step 2
Channel 1 2 3 4 5 6 7 8 9 10 11
f
AP 70 84 80 87 90 83 91 98 89 77 60
g
Sta1 80 88 90 95 93 85 87 91 95 82 77
Sta2 50 70 60 90 88 79 87 92 94 90 87
…. …. …. …. …. …. …. …. …. …. …. ….
Stak …. …. …. …. …. …. …. …. …. …. ….
and are channel condition evaluation functions
Step 3
Dynamic Spectrum Access
Algorithm
The interference index is calculated as a weighted sum of each node’s IS. The AP takes half the weight and each station shares equally for the other half Interference Index 12:00~12:05
Channel
Step 3
©
(j)
Interference Index
1
71
2
85
3
83
4
87
5
90
6
79
7
88
8
95
9
94
10
89
11
68
N
Interference Index j
= ISj(Ap) / 2 +
Selected Channel :
Σ ( ISj(Stak) / (2*N)
k=1
N
*
J = argmax { ISj(Ap) / 2 + j
Σ ( IS (Stak) / (2*N) } k=1 j
The above weight scheme is designed for fairness. It can be changed for other special requirements, such as QoS; In this case, each station may have different weight according to its request for service
Dynamic Spectrum Access
Channel Selection Algorithm Vs Random Channel Selection
Experiment was executed between CRC building 2and CSC building in August 11th, 2011
©
Future Work An improved multi-radio, multi-band WiFi_CR is in the works. Cognitive Femtocells Sub-700 MHz WiFi_CR for TVWS applications. Increasing collaborations: the more, the merrier…putting a practical spin on CR in the ISM environment. Moving into IEEE 802.11n; LTE, and beyond.
©
©
Thank you….Questions?
![[PDF] Cognitive Radio Networking in the ISM Band - Free Download PDF (2024)](data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mP8/h8AAvMB+NzmkbcAAAAASUVORK5CYII=)