EnergyDetection UWB Receivers with Multiplel Energy Measurements E
Energy-Detection UWB Receivers with Multiplel Energy Measurements E. Arias-de-Reyna Department of SP and Communications University of Seville Spain A. A. D’Amico and U. Mengali Department of Information Engineering University of Pisa Italy UWB 4 SN 2005, Lausanne
Outline ü Conventional Energy-Detection (ED) Receiver ü Comparisons with other schemes ü Improved ED Receiver ü Performance evaluation ü Conclusions
Optimal decision strategy (1/2) a K-1 =0 a K =1 a K+1=1 noise t Assumptions ü Channel Response (CR) is unknown ü Perfect synchronization Problem Joint estimation of CR and data symbols
Solution r 2(t) noise r(t) . 2 r 2(t) t x 0 ak = 0 if x 0 > y 0 1 if x 0 < y 0
Conventional ED Receiver “ 1” “o” UWB channel Tb (. ) 2 Compare synch A. Rabbachin and I. Oppermann, UWBST, 2004 M. Weisenhorn and W. Hirt, UWBST, 2004 C. Carbonelli and U. Mengali, TWC 2005/06
Transmitted reference (TR) +1 -1 +1 D t Ts
Differential transmitted reference (DTR) +1 -1 -1 t Ts
Performance comparisons ED/TR DTR
Complexity comparisons (. ) 2 Compare ED synch TR/DTR D synch
Small energy chips r 2(t) t y x 0 0 ak = 0 if x 0 > y 0 1 if x 0 < y 0 t x 0 ak = x N-1 y y 0 N-1 0 if xn > yn 1 if xn < yn
Side information r 2(t) t t x 0 x N-1 y x s 0 y 0 N-1 y s Problem: Given s , what is the optimal decision strategy based on the observation of s N-1 x and y ?
Optimal decision strategy a k=0 p( x , y |a k=0, s ) p( x , y |a k=1, s ) Optimal a k=1 0 if ak = 1 if s nx n > s ny n s nx n < s ny n Approximation (energy correlation)
Improved ED receiver “o” sk “ 1” UWB channel Tb (. ) 2 Compare synch
Performance (1/3) conventional improved 2 d. B
Performance (2/3)
Performance (3/3) s s k known k estimated
Another approach (Weisenhorn and Hirt, ICUW, Sept. 2005) Problem: Given r(t) profile 2 and the channel average power-delay (t) , which is the ML decision rule? A function of 2 (t) and 0 “o” w(t) “ 1” UWB channel Tb (. ) 2 Compare synch
Performance comparisons ED ED WH method Improved WH method
Estimation of s (1/2) Tb Tb noise x k-th bins k y s k s = E{ x k- yk} k k
Estimation of s (2/2) t x (0) k y (0) k s = E{ x k- yk} k x (1) k y (1) x (N-1) k k 1 s = k N ^ y (N-1) k N-1 K=0 x (n) - y (n) k k
Conclusions Pros: ü Improved performance over conventional ED ü Automatic adaptation to channel statistics Cons: ü Higher complexity (sampling rate)
- Slides: 21