Multi-Carrier Waveforms effect on Non-Relay and Relay Cognitive Radio Based System Performances

Transcription

1 Multi-Carrier Waveforms effect on Non-Relay and Relay Cognitive Radio Based System Performances By Carlos Faouzi Bader and Musbah Shaat Senior Associate Researcher, SIEEE Centre Tecnològic de Telecomunicacions de Catalunya Barcelone, Spain

2 Outlines Overview of MC techniques. A Two-Step Resource Allocation Algorithm in Multicarrier Based Cognitive Radio System System Model Problem Formulation Proposed Algorithm Computational Complexity Simulation Results Conclusions Comparison of OFDM and FBMC Performance in DF Multi-Relay Cognitive Radio System Model Problem Formulation Optimal Solution Suboptimal Solution Complexity Comparison Simulation Results Conclusions 2

3 Overview of MC techniques. 3

4 Overview of MC techniques. Multicarrier Communications (MC) are promising techniques in Cognitive Radio CR Moreover, MC based CR systems can meet the spectrum shape requirements by deactivating (i.e. nulling) the subcarriers where the PU is currently transmitting or the subcarriers that can potentially interfere with other users. The multicarrier systems offers very flexible multiple access and spectral allocation of the available spectrum which can be performed in the CR system without any extra hardware complexity ( really?). Several parameters can be adjusted in the system like subcarrier spacing, subcarrier number, modulation, coding and powers. 4

5 Overview of MC techniques, Cont. OFDM is the most common MC technique. WLAN, WMAN, etc. IEEE Large side lobes, and use Cyclic Prefix (CP)! 5

6 Overview of MC techniques, Cont. Filter Based MC scheme: Filter bank can be defined generally as an array of N filters that processes N (different or equal) input signals to produce N outputs. If the inputs of these N filters are connected together, the system -in analogous manner- can be seen as analyser to the input signal based on each filter characteristics. Therefore, this type of filter bank is called analysis filter bank (AFB). On the other hand, by adding the outputs of the filter array, a new signal is synthesized and hence this type of filter bank is called synthesis filter bank (SFB) 6

7 Overview of MC techniques, Cont. The FBMC systems are classified into three main methods: 1. Offset quadrature amplitude modulated OFDM (OQAM-OFDM) by [Chang], [Siohan] 2. Cosine-modulated multitune (CMT) 3. Filtered multitune (FMT). -The isotropic orthogonal transfer algorithm (IOTA) has been presented developed [Siohan] in order to adapt the FBMC system to match the channel time and frequency dispersion. -CMT is considered by the under-development IEEE P1901 standard for power line communication (PLC) systems. Different from OQAM-OFDM and CMT systems, FMT does not allow the subcarrier overlapping [Amini]. [Chang] RW Chang, High-speed multichannel data transmission with bandlimited orthogonal signals, Bell sys. Tech. J, vol. 45, pp , [Siohan] P. Siohan and C. Roche, Cosine-modulated Filterbanks based on extended Gaussian Functions, IEEE Transactions on Signal Processing, vol. 48, no. 11, pp , [Amini] A. Amini, R. Kempter, and B. Farhang-Boroujeny, A comparison of alternative filterbank multicarrier methods in cognitive radios, in 2006 Software Defined Radio Technical Conference and Product Exhibition (SDR 06), Orlando-Florida, Nov

8 PHYDYAS Prototype Filter Extended Gaussian Function (EGF) Overview of MC techniques, Cont. P. Siohan and C. Roche, Cosine-Modulated Filterbanks based on Extended Gaussian Functions, IEEE Trans. Signal Processing, vol. 48, no. 11, pp , Nov M.G. Bellanger, Specification and Design of a Prototype Filter for Filter Bank based Multicarrier Transmission, in IEEE Int. Conf. on Acoustics, Speech, and Signal Processing, Proceedings. (ICASSP 01)., May 2001, vol. 4, pp

9 Overview of MC techniques, Cont. OFDM is the most common MC technique. 1- Large side lobes! high mutual interference. 2- Cyclic Prefix (CP)! reduce the spectral efficiency. FBMC can overcome the OFDM limitations. 1- High spectral containment signal! reduce the side lobes. 2- No CP extension! increase the spectral efficiency. 3- More robustness to the time and frequency offsets. Adrian Kliks, Hanna Bogucka, Faouzi Bader, Musbah Shaat, Oliver Holland, FBMC and GMC Capabilities for TV White Space and Cognitive Radio, IEEE Working Group Meeting, Osaka, Japan, March Paweł Kryszkiewicz, Hanna Bogucka, Adrian Kliks, Oliver Holland, Faouzi Bader, Spectrum Shaping Algorithms for Cognitive Radio, IEEE Working Group Meeting, Osaka, Japan, March

10 Resource Allocation Algorithm in Multicarrier Based Cognitive Radio System 10

11 Outlines Overview of MC techniques. A Two-Step Resource Allocation Algorithm in Multicarrier Based Cognitive Radio System System Model Problem Formulation Proposed Algorithm Computational Complexity Simulation Results Conclusions Comparison of OFDM and FBMC Performance in DF Multi-Relay Cognitive Radio System Model Problem Formulation Optimal Solution Suboptimal Solution Complexity Comparison Simulation Results Conclusions 11

12 Introduction The spectrum scarcity is considered as one of the serious problem. Cognitive radio (CR) aims to increase the spectrum utilization. Multicarrier communication has been suggested as a candidate for cognitive radio (CR) systems. OFDM suffers from high interference to the primary user (PU) and the cyclic prefix insertion decreases the system capacity. FBMC doesn t need any CP insertion and can overcome the interference problem. 12

13 System Model CR base station (CBS) Primary system base station (SU) Primary User (PU) (PU) Secondary User (SU) The CR system coexist with the PU s radio in the same geographical location. The downlink scenario will be considered. Each of the two system causes interference to each other. 13

14 System Model, Cont. B 1 B 2 B L Active PU 1 band Non- Active band Active PU 2 band Active PU L band Frerquency 1 2. f N The SU and PU band are exist side by side. Mutual interference is a limiting factor affect the performance of both systems. The CR can use active and non-active bands. The introduced interference to band should be below I T B l l th th l where l T th l th is the interference temperature limit. 14

15 System Model, Cont. th The interference introduced by the i subcarrier to PU band is th the integration of the PSD of the i subcarrier across the PU band dib 2 2, I d P g f df P i i i i i i i dib 2 The PSD expression depends on the use multicarrier technique. 15

16 Problem Formulation The objective is to maximize the total capacity of the CR system subject to the interference introduced to the PU s and total power constraints. 2 M N Pi, m h i, m P1: max im, log P i m=1 i=1 i Subject to im, M m=1 M im, N m=1 i=1 m=1 i=1 0,1, im, 1, i i, m i, m T P 0, i 1, 2,, N i M N P P P I i, m i i th 16

17 System Model, Cont. The problem is combinatorial optimization problem. The complexity grows with the input size The problem is solved in two steps by many of suboptimal algorithm 1. Subcarriers are assigned to the users. 2. Power allocated to the different subcarriers (virtually as single user multicarrier system) Its proofed that the maximum data rate in downlink can be obtained if the subcarriers are assigned to the user with the best channel. P N i i 2 : max log2 1 P 2 i i1 subject to N i1 N i1 P I i i th P P, P 0, i 1, 2,..., N i T i Ph SCEE Technical Seminar, SUPÉLEC, Rennes, France. December 6,

18 Problem Formulation, Cont. The problem is convex, by using the Lagrange optimization we can get P * i 1 i 2 h i 2 Solving more than one Langrangian multipliers is computational complex. Ellipsoid or interior point methods can be used with complexity Computationally efficient algorithm is needed for practical applications. O N 3 18

19 Proposed Algorithm, Cont. For any set of subcarriers, Optimal solution of the optimization subject to total power constraint only is waterfilling. Optimal solution of the optimization problem subject to interference only can be solved by means of Langrangian and given by P '( Int) i ( Int) l 2 1 ( Int) 2 l i hi Nl 2 i Ith 2 in h l i 19

20 Proposed Algorithm, Cont. Assume that each subcarrier is belonging to the closest PU band and only introducing interference to it. The effect on the net interference will seen after within the simulation. B 1 B 2 f Non-Active band Active PU 1 band Non-Active band Active PU 2 band Non-Active band Frequency 1 2. N N 1 N 2

21 Proposed Algorithm, Cont. For any set of subcarriers N l, Optimal solution of the optimization subject to total power constraint only is waterfilling. Optimal solution of the optimization problem subject to interference only can be solved by means of Langrangian and given by 2 '( Int) 1 Pi ( Int) 2 ( Int) l I th l i i N l l i h 2 in h i 2

22 Proposed Algorithm, Cont. Assign the maximum power that can be allocated to each subcarrier by optimization subject to interference constraint only Power Pmax Pmax Subcarriers N1 N2

23 Proposed Algorithm, Cont. Test the total power constraint, if it is satisfied the solution is found else continuo. Power Pmax Pmax Subcarriers N1 N2

24 Proposed Algorithm, Cont. Distribute the available power according to the waterfilling with the constraint that the power allocated to each subcarrier never exceeds P i max. Power Pmax Pmax Subcarriers N1 N2

25 Proposed Algorithm, Cont. The power constraint is satisfied with equality, which is not the case for the interference constraint. Determine the left available interference and the set of the subcarriers that reach the maximum. A l Power Source of the left interfernce Source of the left interfernce Subcarriers Set A1 N1 Set A2 N2

26 Proposed Algorithm, Cont. Update the values of Pimax in the set according to the left interference and perform the waterfilling again. The power and interference constraints are approximately satisfied with equality. A l Power Initial power loading Final power loading Updated Pimax Updated Pimax Subcarriers Set A1 Set A2 N1 N2

27 Computational Complexity The overall complexity of the algorithm is lower than Where is the number of waterfilling iterations. is estimated via simulations Average Maximum N max log O N N N O L average , i.e. 0,5

28 Simulation Results N=32 subcarriers, M=3 SU s; N1=N2=16, PT=1 watt, L=2 PU s. B 1 B 2 f Non-Active band Active PU 1 band Non-Active band Active PU 2 band Non-Active band Frequency 1 2. N N 1 N 2 Y. Zhang, Resource allocation for OFDM-Based cognitive radio systems, Ph.D. dissertation, Univ. of British Columbia, Vancouver, December 2008.

29 Capacity (Bit/Hz/sec) Capacity (Bit/Hz/sec) Simulation Results N=32 subcarriers, M=3 SU s; N1=N2=16, PT=1 watt, L=2 PU s Optimal-OFDM PI-OFDM Zhang-OFDM Optimal-FBMC PI-FBMC Zhang-FBMC Optimal-OFDM PI-OFDM Zhang-OFDM Optimal-FBMC PI-FBMC Zhang-FBMC Ith-Watt Ith-Watt x 10-5 Low Interference Y. Zhang, Resource allocation for OFDM-Based cognitive radio systems, Ph.D. dissertation, Univ. of British Columbia, Vancouver, December 2008.

30 Net Interference (Ith1) Net Interference (Ith2) Simulation Results, Cont. Net Interference Vs. Interference constraint PI-OFDM Zhang-OFDM PI-FBMC Zhang-FBMC PI-OFDM Zhang-OFDM PI-FBMC Zhang-FBMC Threshold (Ith1) Threshold (Ith2) M. Shaat, F. Bader, Computationally Efficient Power Allocation Algorithm in Multicarrier Based Cognitive Radio Networks: OFDM and FBMC Systems. EURASIP Journal on Signal Processing Advances (ASP). Volume 2010, Article ID March 2010.

31 Conclusions Low complexity sub-optimal resource allocation algorithm for multicarrier based CR networks is presented. The objective was to maximize the total downlink capacity of the CR network while respecting the available power budget and guaranteeing that no excessive interference is caused to the PU s. With a significant reduction in the computational complexity, the proposed PI-algorithm achieves a near optimal performance and outperforms the sub-optimal algorithms proposed so far. The obtained results contribute in recommending the use of FBMC physical layer in the future cognitive radio systems. We are currently working on extension of the proposed scheme by considering more interference constraints as well as channel estimation errors in the cognitive and primary links. 31

32 Comparison of OFDM and FBMC Performance in DF Multi-Relay Cognitive Radio 32

33 Outlines Overview of MC techniques. A Two-Step Resource Allocation Algorithm in Multicarrier Based Cognitive Radio System System Model Problem Formulation Proposed Algorithm Computational Complexity Simulation Results Conclusions Comparison of OFDM and FBMC Performance in DF Multi-Relay Cognitive Radio System Model Problem Formulation Optimal Solution Suboptimal Solution Complexity Comparison Simulation Results Conclusions 33

34 System Model 34

35 System Model, Cont. th The interference introduced by the i subcarrier to PU band is th the integration of the PSD of the i subcarrier across the PU band dib 2 2, I d P g f df P i i i i i i i dib 2 The PSD expression depends on the use multicarrier technique. 35

36 Problem Formulation Rm S D S D 36

37 Problem Formulation The objective is to maximize the throughput subject to the available power budgets and to the interference constraints. Transmission path (relayed/direct), subcarrier matching, relay assignment and power allocation are required to be optimized. 37

38 Problem Formulation. Cont. Rm S D S D 38

39 Problem Formulation. Cont. 39

40 Optimal Solution 40

41 Optimal Solution. Cont. and the dual function is defined as follows M. Shaat and F. Bader, Joint Resource Optimization in Decode and Forward Multi-Relay Cognitive Network with Direct Link, accepted at the IEEE Wireless Communications and Networking Conference (IEEE WCNC'2012). Paris, France. April

42 Optimal Solution, Cont. 42

43 Optimal Solution, Cont. Construct the dual function and initialize the dual variables. Assume that the subcarrier is used for relayed transmission - Find the allocated power for every (j,k,m) assignment. - Substitute the evaluated power and find the best relay m for every (j,k) pairing. Assume that the subcarrier is used for direct transmission - Find the allocated power for every subcarrier j. Determine the cost function for every subcarrier (compare the costs in case of relayed and direct transmission). Solve the linear assignment problem to find the optimal subcarrier pairs and optimal relayed/direct transmission selection. If the stopping criteria is not fulfilled, update the dual variables and perform the steps again. 43

44 Suboptimal Solution, Cont. The dual decomposition approach has high computational complexity. where T is the number of iteration required to converge which is usually large. To reduce the computational complexity, a suboptimal algorithm is proposed to jointly optimize the different system resources. 44

45 Suboptimal Solution, Cont. Musbah Shaat and Faouzi Bader, Asymptotically Optimal Resource Allocation in OFDM-Based Cognitive Networks with Multiple Relays, IEEE Transactions on Wireless Communications, Vol. 11, NO.3, March

46 Complexity Comparison Optimal: Dual decomposition technique. Suboptimal: Proposed Algorithm. SNR: Subcarrier and users are assigned based on their SNR values. Random: Subcarrier and users are assigned randomly. For SNR and Random, Powers are evaluated by solving the optimization problem with known and 46

47 Capacity (Bit/Hz/sec) Simulation Results OFDM Optimal Suboptimal SNR Random 4 P S =P Rm =20 dbm P S =P Rm =0 dbm Interference threshold I th (dbm) 47

48 Capacity (Bit/Hz/sec) Simulation Results, Cont. 6 OFDM 5 4 Optimal Suboptimal SNR Random 3 I th =-10 dbm 2 1 Ith=-30 dbm Source and relays power budget P S =P Rm (dbm) 48

49 Simulation Results, Cont. FBMC Figure 3: Achieved capacity vs. interference threshold with P s =P Rm = 0 dbm. OFDM is plotted in solid lines, and FBMC is plotted in dashed lines. 49

50 Simulation Results, Cont. FBMC Figure 3: Achieved capacity vs. interference threshold with P s =P Rm when P Ith. = -30 dbm. OFDM system is plotted in solid lines, FBMC is plotted in dashed lines. 50

51 Conclusions An optimal and Low complexity sub-optimal resource allocation algorithms for cognitive multicarrier based DF cooperative relay network was presented. The objective was to maximize the total downlink capacity of the CR network while respecting the interference and power constraints. With a reduction in the computational complexity, the proposed algorithm achieves a near optimal performance and outperforms the SNR and random based algorithms. The efficiency of using FBMC is shown to be better than OFDM in CR systems. 51

52 Thank you for your attention For any requested information please contact Dr. C. Faouzi Bader, at s: {cbader, 52