1. Efficient QoS for secondary users in cognitive radio systems

2. Introduction

3. Spectral overhead -> for secondary QoS support
Spectrum sensing and secondary link maintenance -> to ensure protection of the PU communication and maintain a given QoS
Spectral overhead -> for secondary QoS support
Trade-off in using overhead (sensing vs. link maintenance)
인지무선:전파 환경을 측정하여 측정된 전파 환경에 적합하게 무선 기기의 운용 파라미터를 설정하여 동작하는 무선 기술. 예를 들어 무선 기기의 전송 용량을 채널 특성에 맞게 최대화, 기기 간 간섭 최소화, 다른 기종 시스템 간에 상호 동작성 촉진, 또는 비사용 주파수를 찾아서 1차 사용자가 사용하지 않는 시간에 이용하는 기술이 모두 이 범주에 속한다.

4. Spectrum sensing Sensing performance metrics Diversity approaches
Periodic sensing organization

5. Initial sensing / Periodic sensing
Initial sensing - Not time-critical - Secondary user group (SUG) forms a secondary user link (SUL) - After the setup of an SUL, periodic sensing has to be done
Periodic sensing - Time-critical - PU band has to be vacated within the maximum interference time ( )
인지무선:전파 환경을 측정하여 측정된 전파 환경에 적합하게 무선 기기의 운용 파라미터를 설정하여 동작하는 무선 기술. 예를 들어 무선 기기의 전송 용량을 채널 특성에 맞게 최대화, 기기 간 간섭 최소화, 다른 기종 시스템 간에 상호 동작성 촉진, 또는 비사용 주파수를 찾아서 1차 사용자가 사용하지 않는 시간에 이용하는 기술이 모두 이 범주에 속한다.

6. Fundamental problem of spectrum sensing
It is impossible to reliably detect the PU in a certain spectrum range while at the same time performing data transmission in that range
Given the constraint, the amount of spectrum used for sensing requires spectral overhead
QoS 향상 목적으로 IEEE e가 새로 제안한 채널 접속 방식은 HCF(Hybrid Coordination Function)이다.  HCF는 EDCF(Enhanced DCF)라는 경쟁 기반의 채널 접속 방식과 폴링 방식(Polled Channel Access)을 함께 사용한다.

7. Spectrum sensing Sensing performance metrics Diversity approaches
Periodic sensing organization

8. Sensing performance metrics
The usually applied measure for reliability -> : the probability of not detecting a PU although it is present -> the sensing process has to ensure to keep below a certain threshold
Performance measure for the sensing process -> : the probability of detecting a PU although it is not present
QoS 향상 목적으로 IEEE e가 새로 제안한 채널 접속 방식은 HCF(Hybrid Coordination Function)이다.  HCF는 EDCF(Enhanced DCF)라는 경쟁 기반의 채널 접속 방식과 폴링 방식(Polled Channel Access)을 함께 사용한다.

9. Spectrum sensing Sensing performance metrics Diversity approaches
Periodic sensing organization

10. Diversity approaches
In order to achieve better reliability and quality of sensing -> reduce the noise of individual measurements by applying diversity
Three possible diversity dimensions: time, frequency, and space
Diversity in time - sensing over a certain time span - sensing time is strictly bounded by

11. Diversity approaches
Diversity in frequency - each subchannel can be used to get a sensing sample - diversity in the frequency domain is strictly limited: bandwidth of the PU band
Diversity in time and frequency directly increases the spectral overhead

12. Diversity approaches
Spatial diversity requires spectral overhead - sensing itself - exchange of the sensing data
The overhead for distributed sensing depends on the number of participating sensors

13. Spectrum sensing Sensing performance metrics Diversity approaches
Periodic sensing organization

14. Periodic sensing organization
Three different approaches - interrupted sending - dynamic frequency hopping (DFH) -> more complex radio front-end is needed - partial sensing -> similar to the DFH

15. Link and QoS management
Link setup, maintenance, and release
Link reconfiguration
Reconfiguration compensation

16. Contiguous and non-contiguous SULs
SUGs create SULs to perform data transmission
Two possibilities for constructing SULs - contiguous and non-contiguous
They assume non-overlapping SULs

17. Link and QoS management
Link setup, maintenance, and release
Link reconfiguration
Reconfiguration compensation

18. Link setup
Setup of an SUL only happens once at the beginning of the communication
This can be done in a centralized or distributed manner
Centralized approach - central controller of the SUG has to gather the sensing results, decide which resources to use for the SUL, and distribute the decision back to all SUG members

19. Link setup
Distributed approach - the members of the SUG jointly decide on which resources to use for the SUL
The initial sensing and negotiation of parameters requires spectral overhead

20. Link maintenance
It is responsible for the proper operation of an SUL during its whole lifetime
Link maintenance can be divided into - link reconfiguration: responsible for the release and adding of spectral resources to the SUL - reconfiguration compensation: responsible for compensating for potential temporal performance degradations of the QoS due to reconfigurations

21. Release
The periodic sensing process stops and no further usage of the resources of the SUL is authorized

22. Link and QoS management
Link setup, maintenance, and release
Link reconfiguration
Reconfiguration compensation

23. Link reconfiguration
The periodic sensing process indicates that a PU appeared within a PU band used for the SUL
The communication peers have to stop data transmission on the subchannels to be vacated
Negotiate / decide which new subchannels to add to the SUL

24. Link and QoS management
Link setup, maintenance, and release
Link reconfiguration
Reconfiguration compensation

25. Redundancy approach
Add X redundant subchannels to the subchannels of the SUL
Apply coding in such a way that the receiver can decode the message, if any out of the + X subchannels are received
The amount of redundancy added to the SUL has a direct influence on the spectral overhead

26. Resource reservation approach
Some backup spectrum is reserved for the SUL
The SUL can then be immediately switched to use the backup spectrum
Additional backup subchannels need to be maintained
The amount of resources reserved for an SUL has an influence on the spectral overhead

27. Link maintenance approaches
Non-contiguous SULs and partial sensing

28. Overall system design consideration
Theoretically available spectrum
Sensed available spectrum
Effectively available spectrum

29. The probability of false positive
SUL has to be maintained even if the sensing process reports a false positive
A high indeed allows for a small sensing spectral overhead but imposes a larger overhead for link maintenance
Reducing , results in a larger overhead required for sensing but also in a smaller overhead required for link maintenance

30. Overall system design consideration
Theoretically available spectrum
Sensed available spectrum
Effectively available spectrum

31. Theoretically available spectrum
The theoretically available spectrum -> the sum of spectrum that is not used by the PU
It can be seen as a benchmark to compare the performance of different DSA approaches
Up to now, there are only very limited publicly available research results analyzing

32. Overall system design consideration
Theoretically available spectrum
Sensed available spectrum
Effectively available spectrum

33. Sensed available spectrum
is the spectrum sensed to be available for secondary usage
How close this estimate comes to the theoretically available spectrum depends on the quality of the sensing process -> depends on the probability of false positives
does not consider the spectral overhead required for sensing

34. Overall system design consideration
Theoretically available spectrum
Sensed available spectrum
Effectively available spectrum

35. Effectively available spectrum
is the part of the spectrum that actually can be used for secondary data transmission according to the spectrum sensing results achieved on the basis of the used approach for spectrum sensing and SUL reconfiguration

36. System model Performance analysis Performance results
System design example
System model
Performance analysis
Performance results

37. System model
They assume for their investigation a system based on OFDM -> spectrum sensing can be done in parallel
They consider a system model - partial sensing - non-contiguous SULs

38. System model
different PUs each covering a bandwidth of B hertz -> it divided into subchannels of bandwidth
: the probability that a PU is active within
= the probability of detecting the PU

39. System model per slot can be computed as =1-
depends on the number of sensing samples (N)
N depends on and
The system based on OFDM -> there are subchannels for sensing
N = * *

40. System model
= 100, = 50, = 10, = 0.2, = 0.99, = 500 kHz, = 0.5 s and = 0.1 s
They use these values for their analysis

41. System model Performance analysis Performance results
System design example
System model
Performance analysis
Performance results

42. Performance analysis
The effectively available spectrum only considering sensing overhead ->
The probability that a subchannel has to be replaced ->

43. Performance analysis
The probability that there are not enough subchannels available to support the required QoS ->
Choosing a target probability of =0.01 they can find
Using , can be determined ->

44. System model Performance analysis Performance results
System design example
System model
Performance analysis
Performance results

45. Performance results

46. Conclusions

47. Conclusions They have investigated spectral overhead
They show that there is trade-off between the spectral overhead used for sensing and the spectral overhead used for link maintenance
Their work was restricted to the case of non-overlapping SULs