You are on page 1of 14
CORRECTION FACTORS FOR FRESNEL KNIFE- EDGE DIFFRACTION USING UTD JF. Legendre and G. Chassay L.C.S.T. LN.S.A. 20, avenue des buttes de Céesmes 35043 Rennes, France. Colas des Franes Centre d'Electronique de L'Armement (CELAR) 35170 Bruz, France. KEY TERMS Knife-edge diffraction, Field prediction, Uniform Theory of Diffraction (UTD). ABSTRACT In this present paper, shortcomings of the Fresnel knife-edge theory are discussed and simple analytical correction factors are obtained with the help of UTD single wedge results. 1. INTRODUCTION Different technics [1] based on Fresnel knife-edge diffraction theory have been applied to the problem of predicting propagation loss on short radiolinks, where numerical terrain models are available.The more notable are Deygout method, Bullington method and Epstein-Peterson method. However, they showed up some shortcomings, compared with more recent prediction propagation models which use UTD. They can lead sometimes to poor prediction accuracy because such important parameters as polarization, nature of the the ground or spherical illumnation are not taken in account, First, we apply the knife-edge theory and the UTD for a single plane edge, and then, we can deduce some corrections factors of the first theory which enable us to overcome its shortcomings. 2. DERIVATION OF THE DIFFRACTED FIELD BY A SINGLE PLANE EDGE 2.1.The Fresnel knife-edge formulation We consider a straight edge screen (perfectly absorbing) which is interposed between the transmitter E and the receiver R, as, shown on figure I. The heigth h is the distance between the edge and the direct path ; its value is positive when the transmitter and the receiver are in sigth, and negative in the other case. The fieldstrength (normalized with respect to the LOS field Eo) at point R is determined as the sum of all contributions from the secondary Huygens sources in the plane above the edge. Thus yields the well known Fresnel integral as eal oni. Pa a) with Uo= hy 2k ot isa (2) Now, we define the following integral g(x) woo | vn 1x then, we can rewrite eq. (1) as (4) which shows the contributions of the direct path and the diffracted path. Listen) + siamU,). £09) @) . The UTD formulation The UTD plane wegde diffraction [2] is based on a coefficient D**=d°Fd*which is the sum of two complex functions involving a Fresnel integral g(x) ('s' is referred to the soft polarization, and ‘h’ to the hard polarization). di =-4L g(x) es sign(m- (£9) -) 6) with x= 2kL cos{ $ oe } © For a isotrope point source which irradiates spherical waves, the distance parameter L is expressed as L=SS a Factorazing diffraction coefficient by dr, it yields D“ =d[1¥d*/d]=¢.ws (8) Now, we can apply directly the UTD which sums the contibution of the direct ray and the diffracted ray. Employing (5) and defining 6 = si + s* - sas the diffracted ray path delay, it can be shown that the expression of the normalized fieldstrengh in the UTD format is : Bt _ 1 -sign(x+9-9') Eo 2 awe 260) expl jx--jk8 ) + sign(n+9-9'). @) Figure 1 Geometry of the problem 2.3 Application and comparison If points E and R are far away from the edge and the grazing angle is small (<12°), this result reduces to the Fresnel knife- edge theory. For instance, we suppose that E and R are 1000 m far from the screen, with the same heigth, and the parameter h ranges frome -100 to 100 m. we show on fig. 2 the predicted field at 100 MHz obtained with these two methods. The differences between eq. (4) and (9) in both polarization are given on fig. 3 ; these are less than 1 dB. If the distances transmitter-edge and edge-receiver is set at 100 m, figures 4 and 5 indicates that great differences (few dB) between UTD and Fresnel theory are found, especially in the shadowed region of the screen. This is due to the facts that Fresnel Theory doesn't take in account the polarization and that the incident plane wave assumption is not thru here. 0 = z 2 & 4-10 s 3 a -20 1 - UTD (hard) 2- UTD (soft) 3 - knife-edge -30 -100 Height h (m) 100 Eo! Figure 2 Path loss attenuation (E4) against height h Ls with dg = dy = 1000 m, frequency = 100 MHz = Zz os g 1 5 & = 3 00 t g Sc — s | & 05 - 7 1 - UTD (hard) - knife-edge 2- UTD (soft) - knife-edge “10 -100 Heigth h (m) 100 Figure 3 Path loss difference ( eq.4 minus eq.9) against height h with dy = dg = 1000 m,, frequency = 100 MHz 10 0. 710 @ -20 e a ss Ss 1 «* 3 1 - UTD (hard) -40 2- UTD (soft) 3 3 - knife-edge -50 + -100 Height h (m) Figure 4 Path loss attenuation (BE), against height h with dg = dp = 100 m, frequency = 100 MHz g = g 5 5 = 7s B Se — = Ra 1 - UTD (hard) - knife-edge 2- UTD (soft) - knife-edge T T -100 Heigth h (m) 100 @.5 Path loss difference ( eq.4 minus eq.9) against height h with ds = dg = 100 m , frequency = 100 MHz 3. THE CORRECTION FACTORS 3.1. The perfectly conducting knife-edge The precedent shortcomings of Fresnel knife-edge theory can be overcomed in great deal simply by multiplying the Fresnel integral involved in (4) by a correction factor 2. Comparing the equations (4) and (9), we find easily that Q is the product of three functions (Q =u. v . ws) which are u=so/(si+st) (10) v=elix-iks) an weal ¥ gtr). get) (12) The diffraction parameter Uo defined by (2) is extended to any regions by setting Uo? = x-. The function u is a phase correction factor which is useful only in the illuminated region. The function v takes in account the spherical illumination irradiated from the transmitter E. The parameter w is a correction factor which brings the polarization information ; it permits to correct the wrong Fresnel theory behaviour in the shadowed region of the screen. All these functions converges to unity if the Fresnel knife-edge asumptions are verified. Removed from the ISB optical boundary, we may use the asymptotic form of g(x) for large argument to give (13) Within the ISB region where @-@'=m., the asymptotic form of w st is not exactly equal to unity, as equation (13) does, but it takes the value of Eq. (14) sign(n + 9-9) 4) “ 2 whale nk -cos( which tends to unity if KL is large. Also, in most cases, the very simple formula (13), which is continuous at ISB transition, can be used without computing any Fresenl integral. Now, we can apply the correction factor @ to the second example. The difference of path loss between the UTD and the modified Fresnel Theory is less than 0.25 dB, as illustrated by figures 6 and 7. The first simulated curve for the hard polarization is characterized by oscillations in the illuminated region because the Fresnel knife-edge theory doesn't sum vectorially the direct and diffracted rays (as done with UTD ray optical method). The maximum path loss difference is porportionnal to the term (14). Here we have k=2/3 and the distance parameter L = 50 meters ,and injecting these values in Eq. (14), we find analyticaly that the maximum path lo difference is about 0.236 dB. This example shows that great improvement can be obtained with the correction factor proposed here From the knowmedge of other UTD diffraction coefficients, this method can be extended to the case of a plane dielectric screen with or without transmission mecanism. 0.3 Hard polarization a zz 2 0 zg = s Bor 2 g vl 3 SS = 00 01 +100 Height h (m) 100 Figure 6 Path loss difference (corrected eq.4 minus eq.9) against height h in hard polarization with dg = dg = 100 m, frequency = 100 MHz 0.1 Soft polarization 3 S 5 Path Loss difference (dB) 2 (00 Height h (m) 100 Figure 7 __ Path loss difference (corrected eq.4 minus eq.9) against height h in soft polarization with dg = dg = 100 m, frequency = 100 MHz 3.2. The dielectric knife-edge Thanks to the dielectric diffraction coefficient of Luebbers [3] which involves the Fresnel reflection coefficient R*, it is apparent from (8) that wshal+R cos( %° (5) 3.2. The transparent dielectric knife-edge If transmission mecanism may be possible through the screen, Burnside derived a diffraction coefficient [4] from which we deduce the following correction factor Rs wehal+ (16) coo( 22} .cons( 22 ToT where T *"is the transmission coefficent. For this case, we have to add the transmitted ray contribution in the recieved fieldstrengh. Hence, Et _ 1 -signUs) , psn 1+ sign.) * 2 2 a) +(T*- 1). sign(U,) . e(U,?) 4. CONCLUSION We have described a method for improving the Fresnel knife- edge diffraction using a simple correction factor deducted from an UTD theorical analysis. Moreover, polarization, nature of the screen and point source illumination are introduced in this theory. REFERENCES 1, LD. Parsons, The mobile Radio propagation channel, Pentech Press, london, 1992, p. 40-52. 2. R.G. Kouyoumjian ‘and P.H. Pathak, "A uniform geometrical theory of diffraction for an edge in a perfectly conducting surface", Proc. IEEE, Vol. 62, No 11, Nov. 1974, p. 1448-1461. 3. _R. Luebbers, "Finite conductivity uniform GTD vs knife edge diffraction in prediction of propagation path loss", EEE Trans. Antennas propagat., Vol. AP-32, No 1, Jan 1984, p. 4. W. Burnside andK. Burgener, "High frequency scattering by a thin lossless dielectric slab", EEE Trans, Antennas propagat., Vol. AP-31, No 1, Jun. 1986, p.

You might also like