Distributed spectrum sensing in unlicensed bands using the VESNA platform. Student: Zoltan Padrah Mentor: doc. dr. Mihael Mohorčič

Transcription

1 Distributed spectrum sensing in unlicensed bands using the VESNA platform Student: Zoltan Padrah Mentor: doc. dr. Mihael Mohorčič

2 Agenda Motivation Theoretical aspects Practical aspects Stand-alone spectrum sensing Distributed spectrum sensing Spectrum sensing testbed Experimental results Conclusions

3 MOTIVATION

4 Motivation Motivation Theoretical aspects Practical aspects Stand-alone spectrum sensing Distributed spectrum sensing Spectrum sensing testbed Experimental results Conclusions Introduction Radio spectrum Regulation Usage Using the radio spectrum more efficiently Approach Reusing radio frequency bands Licensed Unlicensed

5 Introduction Radio spectrum 1 Many systems use it: AM, FM, TV broadcast, GSM, UMTS, WiFi, GPS, satellite Systems need to coexist Avoid disturbance (interference) Radio spectrum regulation Frequency band allocation Each system has its own frequency band image credit: Roke Manor reseach,

6 Frequency band allocation image credit: Roke Manor reseach,

7 Usage of radio spectrum Studies about radio spectrum utilization Left: Cabric et al: Implemenation issues In spectrum sensing Bottom: Valenta et al: Survey in spectrum utilization in Europe 7

8 Usage of radio spectrum Studies about radio spectrum utilization Terminal 1 Left: Cabric et al: Implemenation issues In spectrum sensing Terminal 3 Bottom: Valenta et al: Survey in spectrum utilization in Europe Terminal 2 8

9 Usage of radio spectrum Studies about radio spectrum utilization Terminal 1 Left: Cabric et al: Implemenation issues In spectrum sensing Terminal 3 Bottom: Valenta et al: Survey in spectrum utilization in Europe Terminal 2 Terminal 4 9

10 Approach Get information about radio spectrum Take decision on the used frequency band 10

11 Approach Perform database lookup Get information about radio spectrum Perform sensing with a radio Take decision on the used frequency band 11

12 Reusing radio spectrum In licensed bands Examples: TV VHF, UHF, GSM bands Primary user(s) Secondary user(s) Dynamic spectrum access (DSA) In unlicensed bands Examples: ISM bands (868 MHz; 2.4 GHz) Multiple equally threated users Spectrum Sharing (SP) 12

13 Reusing radio spectrum In licensed bands Examples: TV VHF, UHF, GSM bands Primary user(s) Secondary user(s) Dynamic spectrum access (DSA) In unlicensed bands Examples: ISM bands (868 MHz; 2.4 GHz) Multiple equally threated users Spectrum Sharing (SP)

14 THEORETICAL ASPECTS

15 Theoretical aspects Motivation Theoretical aspects Practical aspects Stand-alone spectrum sensing Distributed spectrum sensing Spectrum sensing testbed Experimental results Conclusions Problem formulation Goals Hidden terminal and exposed terminal situations Spectrum sensing Energy detection

16 Problem formulation For solving the artificial spectrum scarcity problem, it is necessary: Experimental-driven research Experimental validation and improvement of sensing algorithms Testbed is needed We assume that either: a) a radio communication experiment is prepared in an ISM radio frequency band b) the radio activity in an ISM band is of interest at a given location In both cases external interference might be observed

17 Goals Defining the system architecture for a testbed Developing software that allows performing spectrum sensing with the VESNA platform Spectrum sensing: Calibration of multiple VESNA devices Evaluation of their performance Performing experiments with them Implementation of the functionalities needed for Integrating multiple VESNA devices in a testbed Communication system of the testbed, supporting experiments Experimental evaluation of the performance of a VESNAbased spectrum sensing testbed

18 Hidden terminal and exposed terminal situations Idea: use multiple radios for observation Each radio performs partial detection Results are centralized Resolves the problems: Hidden transceiver Hidden receiver Relies on other methods for partial detection

19 Spectrum sensing Detecting other radios Spectrum sensing methods Energy detection Eigenvalue based detection Cyclostationary feature detection Matched filter detection Collaborative sensing

20 Energy detection Idea: measure the energy in frequency band and compare it to a threshold Simple to implement Needs correct threshold value: noise floor Does not work well with spread spectrum signals

21 PRACTICAL ASPECTS

22 Practical aspects Motivation Theoretical aspects Practical aspects Stand-alone spectrum sensing Distributed spectrum sensing Spectrum sensing testbed Experimental results Conclusions Used devices VESNA platform Spectrum sensing framework

23 Used devices Sensor network based testbed VESNA platform Low-cost, low-complexity CC1101 radio 868 MHz ISM band CC2500 radio 2.4 GHz ISM band The radios can only provide RSSI values Only energy detection is possible

24 VESNA platform Developed at Jozef Stefan Institute ST ARM Cortex-M3, 64 MHz JTAG, USB, USART PC interface I2C, SPI, PWM, ADC, DAC, USART sensor and actuator interfaces Code library: C/C++ (GCC) MHz, 2.4 GHz radio interface (all ISM bands); TI CC1101, TI CC2500 Software tools: Open Source Eclipse IDE Tool-chain: GNU Compiler Collection Cygwin, Linux environment for Windows JTAG server: OpenOCD JTAG hardware interface: Olimex ARM-USB-OCD

25 VESNA platform Performance: - Comparable to other sensor node platforms, like TelosB or Sensinode - Lot less processing power than a PC Developed at Jozef Stefan Institute ST ARM Cortex-M3, 64 MHz JTAG, USB, USART PC interface I2C, SPI, PWM, ADC, DAC, USART sensor and actuator interfaces Code library: C/C++ (GCC) MHz, 2.4 GHz radio interface (all ISM bands); TI CC1101, TI CC2500 Software tools: Open Source Eclipse IDE Tool-chain: GNU Compiler Collection Cygwin, Linux environment for Windows JTAG server: OpenOCD JTAG hardware interface: Olimex ARM-USB-OCD

26 Spectrum sensing framework Control system Radio VESNA Communication and control On-line processing Communication interface Data storage Off-line processing

27 STANDALONE SPECTRUM SENSING

28 Standalone spectrum sensing Motivation Theoretical aspects Practical aspects Stand-alone spectrum sensing Distributed spectrum sensing Spectrum sensing testbed Experimental results Conclusions Goals Experimental setup Calibration results CC2500 CC

29 VESNA Implementation of spectrum sensing functionality Calibration of the prototype

30 Experimental setup Signal generator Coaxial Cable VESNA Generated signal level Measured signal level Offset value

31 Calibration CC2500 Absolute error: < 6 db Nonlinearity: < 2 db

32 Calibration CC1101 Absolute error: < 8 db Nonlinearity: < 0.5 db

33 Calibration CC1101 Malfunction

34 DISTRIBUTED SPECTRUM SENSING

35 Distributed spectrum sensing Motivation Theoretical aspects Practical aspects Stand-alone spectrum sensing Distributed spectrum sensing Spectrum sensing testbed Experimental results Conclusions Goals Demonstration Devices Environment Representative results Device comparison Introduction Environment Results

36 Goals Demonstrate the functioning of heterogeneous sensing system Benchmark Devices Combinations of devices

37 Demonstration - devices ez430-rf2500 Texas Instruments wireless development tool MSP430 CPU CC2500 radio USRP2 Universal Software Radio Peripheral SBX daugthterboard Software defined radio device GNU radio software VESNA CC2500 radio

38 Demonstration - environment

39 Representative results

40 Device comparison Path loss model with parameters Measurement results from devices Fitting Parameter values Error relative to the model For each device Comparison

41 Device comparison Path loss model with parameters Measurement results from devices Fitting Parameter values Error relative to the model For each device Comparison

42 Device comparison Seminar II Path loss model with parameters Parameter values Fitting Measurement results from devices Error relative to the model One static continuous transmission Multiple measurement locations For each device Comparison

43 Device comparison Seminar II Path loss model with parameters Parameter values Fitting Measurement results from devices Error relative to the model One static continuous transmission Multiple measurement locations For each device Comparison Mean Squared Error (MSE): average of squared error values for each data point 44

44 Environment

45 Results - plotted

46 Results - numerical

47 SPECTRUM SENSING TESTBED

48 Spectrum sensing testbed Motivation Theoretical aspects Practical aspects Stand-alone spectrum sensing Distributed spectrum sensing Spectrum sensing testbed Experimental results Conclusions Architecture Goals Requirements Constraints Measurements Setup Representative results

49 Architecture

50 Architecture Functionality abstracted in resources RESTful design: GET and POST requests All nodes addressable Requests initiated by management and control part

51 Architecture Custom application layer protocol Similar to HTTP

52 Architecture Management and control part Access control HTTP interface Scriptable

53 Goals Everything configurable remotely No physical access Unified control interface Simple design and usage Centralized control and data collection Simplicity, reliability Possibility of easily adding functionality in the future

54 Requirements Spectrum sensing data collection Performance level Nodes Control system Reprogramming functionality firmware image transmission performance level Control system Nodes Reliability

55 Constraints Availability of Internet access for the gateway node Location of light poles Power connections to the light poles Radio connectivity Possibilities for experiments

56 Measurements - setup Goal: measuring radio propagation For the control network

57 Measurements representative results

58 EXPERIMENTAL RESULTS

59 Experimental results Motivation Theoretical aspects Practical aspects Stand-alone spectrum sensing Distributed spectrum sensing Spectrum sensing testbed Experimental results Conclusions Scenario Radio wave propagation in the testbed Link quality categories Experiment scenario Results

60 Scenario In the industrial zone 2.4 GHz ISM band Emulated behavior Scripted Observed by multiple nodes

61 Radiowave propagation

62 Link quality categories 1) 2) 3) 1) Good link quality 2) Medium link quality 3) Bad link quality

63 Experimental scenario

64 Node roles in the experiment (c) (n) Node 17: terminal with cognitive radio capabilities (c) Node 2: terminal without cognitive radio capabilities (n) Rest of the nodes: observers

65 Results Node

66 Results Node

67 Results Node

68 CONCLUSIONS

69 Conclusions (1) Spectrum sensing: energy detection is suitable for low-complexity platform Stand-alone spectrum sensing prototype Developed Calibrated Integrated in a heterogeneous system Accuracy has been determined

70 Conclusions (2) Spectrum sensing testbed Architecture defined Network planning performed Developed, set up Including HTTP like protocol Spectrum sensing experiment Prepared Performed

71 THANK YOU FOR YOUR ATTENTION! Questions?