Breaking Spectrum Gridlock With Cognitive Radios: An Information Theoretic Perspective

Transcription

1 Breaking Spectrum Gridlock With Cognitive Radios: An Information Theoretic Perspective Naroa Zurutuza - EE360 Winter 2014

2 Introduction Cognitive Radio: Wireless communication system that intelligently utilizes any available side information about the activity, channel conditions, codebooks, messages or other nodes with which it shares spectrum. Motivation: spectrum scarcity

3 PSD Interference Threshold Cognitive Radio Paradigms Primary User Overlay CR Underlay CR Interweave CR Freq. Underlay Cognitive users transmit simultaneously with noncognitive users as long as interference is below threshold. Channel Side Information Overlay Cognitive users transmit simultaneously with noncongnitive users by assigning part of their transmit power to assist noncognitive users. Codebook Side Information Interwave Identify available spectrum portions for a specific temporary location and time and use these spectrum holes for cognitive unlicensed transmissions. Activity Side Information

4 Degrees of Freedom Not always possible to obtain capacity region of wireless network: asymptotic characterization using Degrees of Freedom Number of accessible interference-free dimensions At high SNR principal determinant of network capacity CΣ(SNR) = d log (SNR) + o (log (SNR)) d log (SNR)

5 Interference Channel I Goal: to determine capacity region of interference channels Interference channel Rate-splitting

6 Underlay I Channel Side Information required Simultaneous cognitive and noncognitive communication below interference threshold --> Interference temperature Capacity: characterized by translating receive power constraint into transmit power constraint at cognitive transmitter AWGN channel Gaussian MAC channel Average interference constraint --> corresponding transmit power constraint. Gaussian codebooks optimal Constraint in sum of transmit powers Capacity region: union of different MAC with constant sum of powers

7 Underlay II Techniques to fulfill the interference constraint: Multiple antennas: guide cognitive signals away from noncognitive users Spread spectrum / UWB Degrees of freedom: Case of two MIMO users: (M1, N1) and (M2, N2) --> min(m1+m2, N1+N2, max(m1,n2), max(n1,m2)) More than 2 users: at least 1/2 of total degrees of freedom (assuming global channel knowledge)

8 Overlay I Codebook Side Information required: problematic at the start of a communication Cognitive user applies encoding schemes to improve cognitive and noncognitive data rates Simplest overlay CR network:

9 Overlay II Overlay encoding techniques: Rate-splitting: partial interference cancelation at decoders GP binning: precode against intereference Cooperation: relay noncognitive message to cognitive decoders Superposition coding: combination of the above techniques Capacity-achieving regimes: Strong interference Gaussian channel with weak interefernce Common information

10 Overlay III Degrees of Freedom: Two-user channel: 1 (interference channel with no cooperation) or 2 (2x2 MIMO with full cooperation in Tx and Rx) MIMO case (M1, N1) (M2, N2): min{m1 + M2, N1 + N2,(1 1T2)((1 1R2)max(M1,N2) + 1R2(M1 + N2)) + 1T2(M1 + M2),(1 1T1 )((1 1R1 ) max(m2, N1 ) +1R1 (M2 +N1 ))+1T 1 (M1 +M2 )} Cooperation: Relay channel Oblivious cognitive users Aware cognitive users, cooperative noncognitive users Full cooperation and cognition

11 Interwave I Activity Side Information required: noncognitive user detection, sensing Signal degradations, fading, shadowing, dynamic nature of CR etc. Missed detection and False alarms --> ROC SNR Wall Single Cognitive User link: two-switch model

12 Interwave II Opportunistic Channel Selection: Narrow-band techniques: Frequency hopping, Frequency tracking Wide-band techniques: Frequency coding Degrees of freedom: Perfect sensing: 1 --> CΣ(SNR) d log (SNR) Imperfect sensing: depend on duty cycle and false alarm/missed detection probabilities Multiple User Case: Access protocols

13 Conclusions Cognitive Radio: tremendous promise Different paradigms, each require particular Side Information Difficult to provide exact capacity results --> degrees of freedom as metric of sum capacity of cognitive networks Cooperation to improve performance Research on combination of different paradigms