Reconfigurable Intelligent Edges: Illuminating the Shadow Region in Wireless Networks

This paper proposes a new wireless enabling technology for future smart radio environments. The approach aims to enhance signal coverage within the shadow region(s) of wireless networks with the aid of so-called ‘reconfigurable intelligent edges (RIEs)’. RIEs may be installed at the fringes of shadowing objects such as buildings, walls and other obstacles which obscure the optical signal path from a transmitter to a receiver. We investigate two approaches to illuminating the shadow region in wireless networks using RIEs that exploit refraction or diffraction, operating in passive or active mode. The operation of RIE-assisted communications are investigated, in particular the ways they can redirect electromagnetic energy towards regions with little or no wireless network coverage. Following from this, a number of variations of RIEs are tested in real-world scenarios which consider illuminating the shadow region behind high-rise buildings, first in a city center, and then along a shoreline. Refractive RIEs in particular, are shown to provide significant gains compared to the case when no RIEs are involved, enhancing signal reception in the shadow region at street level behind a high-rise building by as much as 12 dB. Summary gain statistics are provided so that they can be easily included in system-level analyses and simulations. Critically, it is shown that reconfigurable intelligent edges offer a low complexity and cost-effective solution for improving connectivity in shadowing-limited environments.


I. INTRODUCTION
into the fabric of urban design. In the process, converting 28 The associate editor coordinating the review of this manuscript and approving it for publication was Meng-Lin Ku . classical network coverage areas into intelligent communi-29 cation and/or sensing environments [1]. 30 Pathloss prediction models and algorithms have long 31 considered the effects of electromagnetic (EM) wave prop-32 agation over buildings [2], [3], [4] and around corners in 33 buildings [5], [6], [7]. Taking a dense city center environment 34 as an example, buildings are often closely spaced and mul-35 tiple stories high. Shadow regions are inevitable due to the 36 high probability of the direct (optical) signal path between 37 the base station (BS) and the wireless user being obscured 38 (or blocked) by high-rise buildings or other obstacles residing 39 in the environment. In these situations, any ensuing network 40 coverage in the shadow region will most likely be the result 41 of diffracted and/or scattered multipath which manages to 42 fitted along the corner 1 of a building to exploit scattering 82 and re-radiate impinging waves towards the shadow region. 83 A multi-point approach, based on a series of repeaters (made 84 up of two antennas and an amplifier), was presented in [18] 85 to improve signal propagation through various stages of an 86 indoor non-LOS wireless link at 2.45 GHz. Unlike [17], the 87 repeaters were placed away from the corner so as to provide 88 LOS links between the transmitter and repeater, and then the 89 repeater and receiver. 90 Complementing these advances, in this work we propose 91 an additional tool aimed at controlling signal propagation in 92 future wireless networks. More specifically, we introduce the 93 concept of a reconfigurable intelligent edge (RIE) that can 94 be used to illuminate shadow regions in wireless networks. 95 As we shall show in the sequel, RIEs can be fitted directly to 96 building edges, are low-complexity and offer a cost-effective 97 method of improving signal coverage in shadowing-limited 98 environments with little or no network coverage. We intro-99 duce two types of RIE which operate by exploiting the 100 physics of refraction and diffraction. Among the different 101 ways that a refractive RIE can be developed include using 102 active or passive frequency selective, band-gap or wideband 103 materials [19], [20], [21], [22], while a diffractive RIE can 104 be implemented using passive or reconfigurable corrugated, 105 periodic or semi-periodic surfaces, developed using metal-106 lic or high conductivity materials [8], [9], [23], [24], [25]. 107 Working within wireless networks, such as 6G, RIEs can 108 be made to work intelligently, for example, by judiciously 109 controlling their excitation, or if supported, adaptively tuning 110 their electrical properties or structure in orchestration with 111 network demands. , the dipole structures were placed at a separation of 0.8 to 1 wavelength from the corner of the building. VOLUME 10, 2022 from the BS towards the shadow region behind the building. 118 In this instance, we consider the RIE to have a 3 4 cylindrical 119 structure, 2 with a radius of r. The RIE, shown in red in 120 Fig. 1(a), is attached to the edge of a concrete building. It is 121 assumed to be capable of redirecting the EM energy arriving 122 from the BS along chosen 'bending' angles θ 1 , θ 2 , . . . , θ N 123 illuminating the shadow region, while the redirectivity is 124 assumed to remain consistent along the y-axis for every 125 angle θ. Using this approach simplifies the 3-dimensional 126 (3D) redirectivity problem to a 2D one and allows a thorough 127 investigation of the functionality of the RIE, discussed further 128 in section III.

129
The system level operation of the RIE involves the com- region. An example RIE-enabled smart radio environment is 133 illustrated in Fig. 1 Fig. 2(a) presents results in terms of the maximum 170 E-field, simulated using the finite element method (FEM) 171 (see [26] and references therein for mathematical formula-172 tions). Before considering the RIE, a perfect electric conduc-173 tor (PEC) cylinder of r = 2/3λ is used to create a benchmark 174 for comparison. As expected, fields of the range ∼ 1 V/m 175 are diffracted along the x-axis, while very little EM energy 176 is diffracted towards the shadow region. When the PEC is 177 replaced by a passive refractive RIE, using a material with 178 known n, a higher average magnitude to E-field can be seen 179 to be directed towards the shadow region. As n increases, the 180 angle of peak directivity from the +x-axis increases, however, 181 there is a limit to the extent that the shadow region can 182 be illuminated. For example, taking n > 1.7, it is observed 183 that a portion of the redirected EM energy begins to get 184 absorbed within the concrete. Considering θ = 0 • aligned 185 to x-axis, across the intermediate bending angles, ranging 186 from 0 • to −20 • ( Fig. 2(b)), the passive refractive RIE with 187 n = 1.41 can be seen to provide relatively uniform coverage. 188 For the more extreme bending angles (e.g. θ < −20 • ) RIEs 189 with higher values of n can be used to provide coverage, 190 however they do this in a more focused manner, i.e. with a 191 reduced angular spread of the signal. Remarkably, the results 192 verify that a passive refractive RIE with a radius as small as 193 2λ 3 is capable of providing enhanced radio coverage within the 194 shadow region. In addition to the examples shown in Fig. 2, 195 and given the refractive nature of the RIE material, it will 196 be shown in later sections that wave-fronts with different 197 (i.e. reconfigurable) angles of incidence on the RIE yield sim-198 ilar coverage within the shadow regions. This suggests that an 199 RIE operating in conjunction with beamforming capable BSs 200 will be a promising solution for expanding radio coverage, 201 even when the RIE is passive.

203
This second class of RIE acts to enhance the diffraction at the 204 building edge. Classical diffraction models such as knife edge 205 or equivalent geometric structure diffraction have shown that 206 small amounts of EM energy are diffracted towards wireless 207 users situated behind single or multiple obstacles [8], [9]. For 208 the first time, we show how to enhance this diffraction with 209 an aid of an RIE. Consider the case where the RIE cylinder 210 shown in Fig. 1 is replaced by a cylindrical numerical PEC 211 of r = 1λ [4], [27], [28], and illuminated by a plane wave 212 impinging at an angle of 45 • from the −x-axis. The goal 213 of an RIE is to keep a portion of the impinging EM energy 214 locked on the surface, and release it towards the shadow 215 region. Theoretically, this can be achieved by converting 216 the smooth PEC surface into a 'bumpy' one, however two 217 additional factors need to be considered. First, the actual area 218 illuminated by any impinging wave is curved, so the standard 219 boundary conditions for a periodic bumpy surface [23] can 220 not be applied directly. This is especially the case when r is 221 comparable to the λ. Secondly, given the curved cylindrical 222   is w (Fig. 3(a)). The surface in Fig. 3 towards the desired shadow region (Fig. 3(b)). However, the same corrugated surface featuring a l of ∼ 0.1λ, approxi-238 mating a so-called rough conducting interface [25] can be 239 seen to enhance the diffraction, as shown in Fig. 3(c). 240 As argued in [29], the majority of the power factor of the 241 surface wave propagating through the corrugated metal sheet 242 is determined by the shape of the uppermost part of the 243 roughness. Hence, a curved corrugated RIE surface, with 244 longer l and a wider gap between the plates, results in low 245 effective capacitance along the outer edge, making it a high 246 impedance surface. When l is reduced, it increases the effec-247 tive capacitance, helping to lock the propagating wave. Thus, 248 in Fig. 3(c), the outer edge of the RIE can be seen to have 249 the highest value of E-field, where surface waves become 250 locked, and are then slowly released towards the shadow 251 region. This phenomenon governs the operation of a passive 252 diffractive RIE.

254
Changing the effective n of a material is an active area of 255 research [19]. Future innovations here will create smart mate-256 rials and surfaces, paving the way for active refractive RIEs 257 which can be retuned in response to requests to illuminate 258 different portions of the shadow region. Under this regime, 259 two different types of aperture can be considered, the first, 260 in which the receiving aperture area is planar, and second, 261 in which receiving aperture area is curved. One approach to 262 creating an active refractive RIE with a planar surface area is 263 to use layers of metasurfaces with periodic or quasi-periodic 264 sub-wavelength structures, capable of redirecting the incident 265 wave towards the desired direction.

266
A number of approaches to create layered structures and 267 their associated active controls are given in [20], [30], 268 and [31]. One such approach is to implement layers with 269 switching capabilities. Each layer is then programmed to 270 slightly tilt the impinging signal, channelling it in the desired 271 direction of propagation. It is worth highlighting that while 272 conceptually relatively straightforward, the practical imple-273 mentation of such an RIE will be non-trivial as a number 274 of boundary conditions between sandwiched layers apply, 275 requiring extensive hardware considerations. Also, due to the 276 number of layers, higher insertion losses are to be expected. 277 To develop an active RIE with a curved surface facing the 278 impinging signal, one option is to divide the RIE cylinder 279 VOLUME 10, 2022  Note that capacitance in a corrugated surface can also be 303 achieved by changing p or even w, while keeping l constant, 304 giving rise to a number of reconfigurability options using tun-305 able electronics implemented on the diffractive RIE surface.

306
One representative RIE is shown in Fig. 4, where three corru-  be seen in Figs. 4 (a) -(c). One way around this is to 313 break the periodicity and create a high impedance along 314 the RIE surface where the incident wave is impinges and 315 decreases the impedance along the corrugated RIE surface. 316 Another way is to create a curved, sinusoidally modulated, 317 impedance surface on the RIE (similar to the planar ver-318 sion in [34]). Nevertheless, the goal of improved signal 319 strength within the shadow region is achievable as shown 320 in Fig. 4(d).

322
Based on our initial analysis, to provide directive coverage 323 within a shadow region, active or passive refractive type 324 RIEs show the most promise (see Fig. 5(a)). This will be 325 advantageous in scenarios where the network knows the 326 approximate location of wireless users and can adapt the 327 'bending' angle accordingly. Correspondingly, to achieve 328 wider coverage across the entire shadow region (and hence 329 low directivity), passive or active diffractive RIEs are a 330 better option ( Fig. 5(b)). Clearly, this form of transmission 331 will be most beneficial in situations where the location 332 of wireless users is unknown or cannot be estimated in 333 advance.

334
Some additional operational considerations are impor-335 tant to note. First, the overall aperture area on the RIEs 336 considered here is likely to be much smaller compared to 337 widely proposed reflecting (e.g. RIS) or refracting surfaces 338 (e.g., transmissive metasurfaces). Hence, the impinging sig-339 nal's power may need to be higher, meaning that for many 340 practical implementations, a focused beam of energy will 341 be required for the excitation of RIE. Secondly, the RIE is 342 expected to maintain its performance within a bandwidth, say 343 for instance, in sub-bands of 5G n73, say 3.46 -3.48 GHz 344 and 3.50 -3.54 GHz (see Fig. 6). However, for wider band- from OpenStreetMaps [35] to replicate a full-scale section of 367 Belfast City in the United Kingdom, as shown in Fig. 7(a). 368 In our system, the RIE is irradiated by a BS situated on 369 the roof of a high-rise building, as depicted in Fig. 1(b).  shoreline (e.g. as found in Miami Beach in the USA and 381 Recife in Brazil, Fig. 7(b)). For this case, and to keep con-382 sistency with the analysis conducted for the city center envi-383 ronment, an 80 m tall building was placed at the edge of a 384 22 m street, as depicted in Fig. 7(d). In contrast to the previous 385 example (Scenario I), there were no buildings on the opposing 386 side of the street. The choice for the distance between the 387 building and the roadside pathway was based on a survey of 388 satellite images from different cities, which were consistently 389 between 22 and 24 m, except for places where hotels with 390 pools are located at the beach. 391 VOLUME 10, 2022 We used the Asymptotic Solver (A-Solver) available within 393 CST Studio Suite to simulate the two example scenarios.

394
In the A-Solver, lossy materials, such as concrete, are not 395 natively supported; therefore we used a lossless dielectric The results obtained from the ray tracing performed in CST 458 were post-processed in Matlab, by converting the simulated 459 E-Field measured by the probes using where P X is the power density measured in Watts/m 2 at each 462 probe, X ∈ {No RIE, Refractive RIE A, Refractive RIE B, 463 Diffractive RIE A, Diffractive RIE B}, and η 0 = 120π is the 464 free-space impedance measured in Ohms.

465
As discussed previously, one of the key challenges in cel-466 lular networks is ensuring sufficient coverage at street level in 467 dense city center environments. This is often hampered by the 468 close proximity of tall buildings, as depicted in Fig. 7, which 469 can shadow signal transmissions from the BS positioned on 470 the top of the high-rise structures. To investigate the ability 471 of RIEs to support BS transmissions in this case, we now 472 employ four of the RIE variations introduced in Section III. 473 In our analysis, we split the street level into three distinct 474 zones as shown in Fig. 8. 4 These are: Zone 1, which con-475 siders the section of the street closest to the building with 476 the RIE; Zone 2, which is made up of the central area of 477 the street in between the two buildings; and Zone 3, which 478 is the section of the street closest to the building on the 479 opposite side of the street.   To conduct our analysis, we make use of the RIE gain 486 calculated at each distance (d), which is defined as

488
where P X and P No IE are defined in (1). 489 We first consider Scenario I. Fig. 9 shows the RIE gain 490 experienced 1 m above street level compared with the dis-491 tance from the base of the building fitted with the RIE.

492
As we can see from the plots ( Fig. 9(a) refractive RIEs A and B was largely identical, with both con-507 sistently offering over 6 dB enhanced gain over the majority 508 of the region compared to the no RIE case ( Fig. 9(a)), which is 509 approximately equivalent to quadrupling the received power. whereas diffractive RIE A only provided a marginal improve-516 ment and diffractive RIE B consistently performed worst 517 overall.

518
These observations are confirmed by the empirical cumula-519 tive distribution functions (CDFs) given in Fig. 10. As can be 520 seen from Fig. 10(a), in Zone 1, both refractive RIEs deliver 521 a signal enhancement over nearly 100% of the shadow region 522 covered by Zone 1 compared to the no RIE case. As shown in 523 Table 1, 5 the mean RIE gains for refractive RIEs A and B in 524 Zone 1 are 6.07 dB and 5.26 dB, respectively. Interestingly, 525 both of the diffractive RIEs are shown to provide an enhance-526 ment over 70% of the shadow region covered by Zone 1, 527 although as discussed above, the RIE gains are much lower 528 than those delivered by the refractive RIEs, suggesting that 529 for this type of scenario, refractive RIEs are the best option 530 for illuminating the shadow region closest to the base of the 531 building. Fig. 10(b) provides the empirical CDFs of the RIE 532 gain in the central region between the two buildings (Zone 2). 533 Here we can see that the refractive RIEs again offer the best 534 performance, with both guaranteeing a signal enhancement 535 for 95% of the shadow region covered by Zone 2 compared 536 to the no RIE case. In Zone 2, the performance of both 537 diffractive RIEs was degraded compared to Zone 1, with both 538 being outperformed by the no RIE case for approximately 539 50% of the zone. In Zone 3, both refractive RIEs performed 540 similar to Zone 1, offering increased RIE gain across the 541 entire zone, with mean RIE gains of 4.90 dB (refractive 542 RIE A) and 5.20 dB (refractive RIE B). Diffractive RIE A 543 outperformed the no RIE case for at least 65% of the shadow 544 region covered by Zone 3, however, as before the benefits 545 here were marginal (mean RIE gain = 1.25 dB, Table 1). 546 In this region, the diffractive RIE B performed poorer than 547 the no RIE case offering a degradation in the mean RIE gain 548 = −1.55 dB (Table 1). For completeness, Fig. 10(d) shows 549 the empirical CDFs of the RIE gains calculated over the entire 550 region between the two buildings (i.e. Zone 1 + Zone 2 + Zone 551 3). With the exception of diffractive RIE B, the majority of the 552 empirical CDF plots reside in the positive half of the figure, 553 providing strong evidence for the use of reconfigurable intel-554 ligent edges to enhance signal coverage in shadowed regions 555 at street level within city center environments. 556 Fig. 11 provides a sample of the empirical CDF plots 557 for the case when the probes are situated: (a) 1.5 m above 558 street level 6 ; (b) at the front face of the 'left' building on 559 the opposite side of the street to the building with the RIE 560 fitted and; (c) at the front face of the 'right' building, i.e. 561 the building with the RIE fitted. Again, in all three cases, the 562 refractive RIEs provide favorable performance, outmatching 563 the no RIE case in the majority of cases. Following a sim-564 ilar trend to results discussed so far, diffractive RIE A per-565 formed best achieving signal enhancements for around 70% 566  to the field strength. In environments such as beach fronts 575 (Fig. 7(b)) or at the peripheries of city centers, the propaga-576 tion geometry may mean that signal components arriving at 577 the receiver are primarily the result of propagation around the 578 building edge or those reflected from the ground. To investi-579 gate this Scenario II, we replicated the simulation performed 580 above, except in this instance, no building was positioned on 581 the opposite side of the street (Fig. 7(d)). Again the street 582 TABLE 1. Mean (Av.) and Standard Deviation (std.) of the RIE Gain in decibels at 1 m elevation from street level in the city center environment (Scenario I). Note maximum and minimum mean RIE gains are highlighted in green and pink respectively.     RIE case across all three zones (Fig. 12(a)). What is most 589 striking, is the significant RIE gain that can be achieved,  (Fig. 12(b)). Despite  center environment, we can see that in nearly all cases (where 623 the optimal refractive RIE was chosen), the RIE gain for 624 the beach front was greater than that achieved in the city 625 center. From Fig. 13, we can see that both diffractive RIEs 626 offered poorer performance than the refractive RIEs, with 627 refractive RIE B performing worst overall. Lastly, and for 628 completeness, Fig. 14 provides the empirical CDFs of the 629 aggregated RIE gain for (a) 1.5 m above street level and; 630 (b) the face of the 'right' building, i.e. the building with 631 the RIE fitted, with the corresponding summary statistics 632 provided in Tables 5 and 6. Here we can see that the refrac-633 tive RIEs again outperform the diffractive RIEs, however 634 it is worth pointing out that diffractive RIE A still man-635 aged to provide a positive RIE gain for over 70% of the 636 shadow regions at the street level and the front face of the 637 building.

639
As we have seen in Section V.B, RIEs have significant poten-640 tial to enhance signal reception in regions behind large shad-641 owing objects such as high-rise buildings. As part of the 642 propagation manipulation process, they will also impact the 643 channel as observed by the receiver, in this case, in the down-644 link. To illustrate how RIEs can alter the channel, we present 645 some snapshots of the characteristics of the downlink signal 646 as experienced in the city center environment for a selection 647 of receiver positions above ground level and along the sides 648 of the buildings. For the study, we focused on time and angle 649 dispersion characteristics which are important in wideband 650 systems (e.g. to understand the potential impact of effects 651 such as intersymbol interference (ISI)) as well as beamform-652 ing systems (to understand the most likely angle of arrival 653 (AOA) and the degree of signal spreading across the angular 654 domain). For brevity, we only consider refractive RIE A in 655 our analysis since it was observed to show the most promis-656 ing results based on the analysis performed in Section V.B. 657 To conduct the channel simulations, we used the same setup 658 as described in Section V.A, however in this instance instead 659 of electric field probes as receivers (or equivalently UEs), 660 we used isotropic receivers to enable antenna coupling cal-661 culations and channel parameters to be computed within 662 the CST A-solver. To obtain the channel snapshots we used 663 7 isotropic receivers for the 1 m above ground level (and again 664 for 1.5 m above ground level) simulations. In this instance, 665 the receivers were placed at the center and boundaries of each 666 zones defined in Section V.A. The receivers along the sides 667 of the buildings were positioned 3 m apart, from ground level 668 upwards.

685
Another observation from Fig. 15 is the different numbers 686 of MPCs which are detected at the UE. For practical purposes, 687 we set the noise threshold at a level of −120 dBm, meaning 688 that any signal components that arrived with a power less 689 than this value were disregarded as it is unlikely they would 690 make a meaningful contribution to the received signal. We can 691 see in Fig. 15(b) that many more signal components arrive 692 at the UE compared to Fig. 15(a). In some cases, MPCs, 693    Fig. 15(b) shows that the use of an RIE 696 in scenarios where UEs are operated close to ground level 697 opens up opportunities for additional signal paths to be cre-698 ated (see discussion of the RMS angle spread below), which 699 may be beneficial for beamforming systems attempting to 700 re-establish broken links by seeking alternative signal paths 701 in the angular domain [37].

702
In wireless systems design, two key metrics for deter-703 mining the likely impact of ISI are the mean excess delay, 704 denotedτ , and the RMS delay spread, denoted σ τ . The mean 705  excess delay may be determined from the PDP as [38] 707 where |α i | represents the amplitude and τ i represents the 708 excess delay of the i th multipath component, respectively.

709
To calculate the RMS delay spread, we evaluate [38] 710

711
where, Table 7 presents the average (av.) and the standard devia- MPCs, enhanced by the RIE, and reflected off the sides of 730 the building contribute significantly to the received power and 731 increase the RMS delay spread. It is also worth highlighting 732 that when used in the city center environment, the RIE causes 733 an increase in the maximum excess delay time compared to 734 the no RIE case. This is presumably due to the enhanced 735 gain offered by the RIE, meaning that signal components that 736 extend below the noise threshold, which would ordinarily 737 not be sensed by the receiver, are now detected. This may 738 or may not be problematic depending on data transmission 739 rates and ISI countermeasures employed such as adaptive 740 equalization [39].

741
Similar to the mean excess delay and RMS delay spread, 742 equivalent metrics can be computed to understand signal 743 dispersion in the angle domain. To characterize the angu-744 lar spread of the MPCs in the city center environment, 745 we computed the mean angle of arrival (AOA),β, according 746 to [38]

751
where, (8) 753 As can be seen from Table 8 AOAs promoted by refractive RIE A (Fig. 17(b)) compared 767 to the no RIE case (Fig. 17(a)) is clearly evident, reinforcing that choosing the right RIE configuration (e.g. refractive 800 index or corrugated surface structure) can lead to consid-801 erable signal enhancements in both environments. Overall, 802 it was found that the RIE gain (that is the improvement 803 compared to the case where no RIE is present) was great-804 est for refractive RIEs. These performed best in practice, 805 offering a mean gain of at least 5.20 dB, with instantaneous 806 gains of up to 12 dB across the shadow region for a user at 807 street level (1 m) in both environments. It is worth remark-808 ing that although we have demonstrated the RIE concept 809 for rooftops in a cellular setting, the same principles apply 810 for manipulating signal propagation around building corners 811 (e.g. to assist device-to-device communications [40]) and by 812 extension within indoor environments to assist with wireless 813 local area network (WLAN) coverage [41]. 814 As with any new technology, many unanswered questions 815 remain. For instance, the successful adoption of RIEs will 816 require a greater knowledge of the relationship between their 817 geometrical and material properties and signal re-directivity. 818 Determining how these structures perform at different fre-819 quencies, and how they can be made sufficiently broadband 820 (e.g. to provide seamless coverage across cellular bands) will 821 be paramount. Understanding how RIEs can be integrated 822 into future smart radio environments will also be important, 823 in particular, how they can be used to augment other prop-824 agation manipulating technologies such as RISs. This will 825 require determining their role in increasing capacity, improv-826 ing reliability and reducing latency. The next stage of this 827 work will include the physical implementation and testing 828 of RIEs using layered surfaces. This will incorporate further 829 detailed channel studies to understand how the inclusion 830 of RIEs impacts the end-to-end spectral efficiency in both 831 outdoor cellular and indoor WLAN settings. 832 diffraction measurements and models at 10, 20, and 26 GHz,'' in Proc. Technology, USA. He is also an Adjunct Professor 1037 at Duke University, USA. His research interests include microwave and 1038 millimeter-wave imaging, multiple-input-multiple-output (MIMO) radars, 1039 wireless power transfer, antennas and propagation, and metamaterials. 1040 He has authored more than 150 peer-reviewed technical journals and con-1041 ference papers, and has been a principal investigator and a co-investigator 1042 on research grants totaling in excess of £10M in these fields. He is a mem-1043 ber of the European Association on Antennas and Propagation (EurAAP). 1044 He was a recipient of several awards, including an Outstanding Postdoc-