Radar-Aided Communication Scheduling Algorithm for 5G and Beyond Networks

Radar and communication coexistence is an upcoming technology with numerous research opportunities in the medium access control (MAC) layer, particularly in scheduling and radio resource management (RRM). More efficient scheduling algorithms are needed with the wide range of applications that the wireless environment is experiencing. We investigate an echo-based scenario in the radar-aided vehicular communication system in which an echo is reflected from a target. Unlike the conventional scheduling mechanisms where signal-to-interference-plus-noise ratio (SINR) is exploited, this paper proposes a new radar-aided communication scheduling algorithm by utilizing parameters such as range and velocity with the classical SINR measurements. The proposed algorithm schedules the available resources by extracting information from the radar echo. The proposed radar-aided communication scheduling scheme provides a more flexible design by adding new parameters, resulting in a more efficient algorithm in a broad variety of scenarios. The proposed scheme is beneficial for B5G communication systems that allow localization and sensing as key features of next-generation wireless networks.

management are not adequately given [20]. The related radio 101 resource management (RRM) studies for radar and commu-102 nication coexistence systems are classified according to the 103 resource management issues, i.e., spectrum sharing, power 104 allocation, and interference management. Literature works 105 have not been done on the radar-aided communication sys-106 tem in the MAC layer to the best of our knowledge [20]. 107 The radio resource allocation and scheduling issues are not 108 thoroughly explored and debated. Different from the existing 109 PHY-based resource management approaches [20], we pro-110 pose a new scheduling algorithm tailored for the radar-aided 111 communication systems by exploiting different parameters 112 that have not been used before, which includes; range and 113 velocity. The proposed solution provides a more flexible 114 scheduling mechanism for the systems' available resources. 115 This paper explains in detail how the estimated parameters 116 can aid RRM, including a novel radar-aided communication 117 scheduling mechanism. 118 The proposed scheduler is evaluated by analyzing its 119 performance with the conventional resource scheduling 120 algorithms. Classical physical resource block scheduling 121 algorithms for the resource allocation are reviewed exhaus-122 tively in [21]. The UE scheduling concept based on fairness 123 and reliability has been extensively studied in the litera-124 ture [22], [23]. PF, RR, and BCQI are the most known 125 scheduling algorithms. Our scheduler output maximizes the 126 system performance of the communication system by max-127 imizing its performance metrics, including communication 128 data rate, spectral efficiency, and system throughput. Higher 129 data rate and spectrum efficiency are required in indoor 130 scenarios [24]. Spectral efficiency should be maximized, for 131 example, if an unmanned aerial vehicle (UAV) covers a high 132 number of targets so the scheduling data can be adequately 133 collected. High system throughput is required due to its     Table 1 shows the list of abbreviations. The remainder 162 of this paper is presented as follows. The system model 163 is explained in Section II. The proposed scheme is illus-164 trated in Section III. Simulation results are discussed 165 FIGURE 2. The system model where the radar information at the BS is leveraged to schedule the resources of multiple mobile users.
in Section IV. Finally, the paper is concluded in 166 Section V. 1 168 We investigate a radar echo-based scenario where a radar 169 echo is reflected from the radar target [20]. In this scenario, 170 the echo is exploited to obtain different scheduling parame-171 ters, such as target(s) range, velocity, and SINR information. 172 We consider a mmWave vehicle-to-infrastructure (V2I) 173 communication system, e.g., supported through 5G cellular 174 network, where mmWave BSs serve as infrastructure for V2I 175 communications. A monostatic radar system is collocated on 176 the BS, which receives the radar echos from the surrounding 177 targets, as illustrated in Figure 2. When a moving target 178 is detected by the radar, the raw echo signal is sent to the 179 communication module at the BS. Side information derived 180 from radar mounted on the infrastructure operating in a given 181 mmWave band is used to schedule the available resources of 182 the vehicular communication system [5]. The objective of the 183 considered system is to schedule the available resources in the 184 BS relying on information extracted from the radar echo.

185
The system under consideration in this study comprises 186 of a BS and multiple mobile users which act as radar tar-187 gets and also as communication receivers. The BS has two 188 major components: (i) A phased array-equipped mmWave 189 communication terminal to communicate with the mobile 190 users, and (ii) a collocated radar system using frequency 191 modulated continuous wave (FMCW) signal. The system and 192 signal models of the communication and radar components 193 are briefly described in the next two subsections [5].   The transmitted signal from the radar frame can be expressed where √ E r is the channel gain of the object, which depends . It should be noted that the investigation of this 244 method is not given in our paper since it is out of the scope of 245 the proposed MAC-based scheduler. After the radar informa-246 tion is extracted, they are fed into the communication module. 247 In the following subsection, the communication model is 248 described.

250
The considered BS employs a mmWave OFDM-based trans-251 mitter with M A antennas which is used to communicate with 252 multiple single-antenna mobile users. Adopting a narrow-253 band channel model with P paths, the channel between the 254 k-th user and the BS can be expressed as [27] where α p denotes the complex gain, a φ p , θ p is the array 257 response vector of the BS, and φ p , θ p represent transmit 258 azimuth and elevation angles of the p-th path at BS. In the 259 downlink, the BS transmits the data symbol s d to the user via 260 the beamforming vector f ∈ C M A . The receive signal at the 261 k-th user can be written as where n ∼ CN 0, σ 2 is the additive white Gaussian noise 264 (AWGN) with a variance of σ 2 , and √ E c is the transmitter 265 gain of the BS. The signal-to-noise ratio (SNR) measured at 266 the k-th user can be written as In the conventional communication channel state infor-276 mation (CSI) acquisition process, a known pilot signal is 277 transmitted, and the communication beams are adjusted based 278 on this signal [28]. The FMCW radar signal being a known 279 signal, is used in the considered system to estimate the 280 CSI for the communication users rather than transmitting 281 additional pilots. As a result, pilot transmission overhead 282 is reduced [26]. Furthermore, the radar measurements are 283 leveraged to optimize the communication scheduler's perfor-284 mance. Based on the classical SINR measurements and the 285 information gathered about the users from the echo signal, 286 namely range (g k ) and velocity (v k ), the BS allocates its 287 available resources to the users. The next section contains 288 more information on the proposed multi-user scheduling 289 scheme. should be high to receive the packet correctly. For example, 320 the minimum required SINR could be 0 dB.

321
The range and velocity of the target can be extracted 322 from the echo signal. The radar receiver uses different signal 323 detection algorithms, e.g. correlation-based methods [30]. 324 Based on the considered coexistence communication and 325 radar use cases, it is found that range and velocity are the 326 critical metrics to be measured. Figure 4 shows the pro-  DL transmission from BS to UE in a typical radar-aided 329 communication system.

330
The scheduling inputs are collected by the echo signal 331 from the target. These inputs include range, velocity, and 332 SINR information. The considered inputs are imported from 333 the considered applications and use cases. Some numerical 334 values of the considered inputs can be shown in Table 2 [37]. 335 It should be noted that the velocity metric, shown in Table 2, 336 refers to the maximal (relative) velocity that should be 337 supported.

338
Based on the considered scheduling inputs, we set proper 339 scheduling rules. The range-based scheduling rule is set 340 where UE near the BS is prioritized due to low path loss. 341 The velocity-based scheduling rule is set where low-speed 342 UEs are prioritized due to their invariant channel conditions. 343 The SINR-based scheduling rule is set where UEs with better 344 SINR are also prioritized due to their high power. We consider 345 a scenario where multiple BSs and UEs exist in the commu-346 nication network. In this case, SINR is measured since the 347 interfering BSs cause interference to the desired UE served 348 by its serving BS [38]. For simplicity, equal radar and com-349 munication transmit powers are assumed.

350
The estimated range and velocity from the considered sys-351 tem, as well as some of the assumptions made in the proposed 352 scheme, are described further. Basically, the mutual distance 353 between k-th UE and b-th BS, d bk (km), can be found as [39] Figure 5 shows the considered multi-cell scenario where 417 the cellular topology is adopted in a single ring hexagonal 418 structure with 7 5G BSs (gNBs), and their positions are at 419 the center of the hexagonal cell in a square area. Users are 420 dropped in the same area according to a uniform spatial dis-421 tribution across the region of interest (ROI), which is around 422 7 × 7 km in size. 423 We focus on downlink transmissions without power con-424 trol, as is the case in the majority of state-of-art proposals. 425 It is assumed that BS knows the DL traffic demands that are 426 cached in its buffer. UEs are associated to BSs based on the 427 strongest average received power, i.e., based on distance, and 428 do not change BS during the simulation. Fading is considered 429 in the numerical simulations in addition to path loss, through a 430 random variable, expressed in dB, distributed as a zero-mean 431 Gaussian with a variance of σ 2 [42].   To understand the functioning of the proposed approach in 442 much more detail, its performance is evaluated by considering 443 the RR, PF, and BCQI methods in terms of average data rate 444 (bits/s), throughput (bits/s), and SE (bits/s/Hz) over 1000 dif-445 ferent channel realizations [43]. 446 The average system and UE data rate, system throughput, 447 and SE are found as follows, respectively: . System data rate of the proposed scheduler compared to the conventional one. The most commonly used scheduling methods are PF, BCQI, 458 and RR in the literature and practice [44]. The available 459 RBs are distributed using the PF, BCQI, and RR. The per-460 formances of PF, RR, and BCQI are analyzed under the 461 proposed approach. For the same simulation scenario, the 462 three scheduling techniques are used.

463
The PF method is one of the most widely used methods for 464 fair scheduling [23], aiming to provide fairness while taking 465 advantage of good channel conditions and dynamically allo-466 cating resources to UEs. It is proven in [45] that the PF is not 467 optimal due to capacity constraints. The RR method provides 468 RBs for UEs without considering channel conditions; this 469 is a simple procedure that ensures fairness [46]. The BCQI 470 method is a common channel-dependent scheduler. The chan-471 nel variations between UEs are exploited in this scheduler to 472 maximize cell throughput at the expense of fairness [47].

474
Here, the simulation results are presented along with their 475 discussions. MATLAB-based SLS is used to evaluate the 476 performance of the proposed approach. The achieved results 477 are shown with the classical methods, i.e., PF, RR, and BCQI. 478 During the simulations, the default parameters are listed in 479  Table 3. The simulation results here showed the realistic 480 case in which the signal experiences interference and noise 481 simultaneously. Because of that, the SINR of UE has been 482 measured when the power of the noise term is zero. Then, 483   the SINR reduces to the signal to interference ratio (SIR).

484
Conversely, zero interference reduces the SINR to the SNR. SINR is only considered. It is demonstrated that the BCQI 492 outperforms the PF and RR in terms of data rate, SE, and 493 system throughput under the proposed scheme. The reason 494 behind this is the channel-dependent characteristics of BCQI 495 while assigning RBs, resulting in superior spectrum utiliza-496 tion. The results also show that the performance of the PF 497 is slightly better than the RR. The RR performs the worst 498 because it reduces throughput, data rate, and SE. On another 499 side, it offers greater resource allocation fairness among dif-500 ferent UEs. It is worth noting that the throughput depends on 501 the simulated channel state, which influences the achievable 502 transmission rate and the UE scheduling mechanism adopted 503 by BS.

505
More effective scheduling algorithms are required to fulfill 506 the needs of new applications such as radar and communica-507 tion coexistence. We propose a new radar-aided scheduling 508 algorithm in a practical vehicular communication scenario 509 by utilizing some parameters, including range, velocity, and 510 the conventional SINR measurements. The fundamental idea 511 is to create the communication scheduler at the BS using 512 the estimation of the scheduling inputs derived from the 513 radar echo signal. The proposed scheme introduces a new 514 degree of freedom to the scheduling design in radar-aided 515 communication systems that leads to an efficient scheduling 516 algorithm. The proposed scheme is evaluated in terms of 517 data rate, UE and system throughput, and SE. The simula-518 tion results prove that the proposed scheduler outperformed 519 the conventional one. Moreover, the obtained results show a 520 significant performance of the proposed scheduler using the 521 BCQI over the PF and RR. The results show that radar can be 522 a valuable source of side information for the communication 523 scheduler over a mmWave V2I link. The proposed algorithm 524 should be viewed as a preliminary solution to demonstrate 525 the approach's practicality, although it can be modified fur-526 ther. As future work, the optimal solutions could be further 527 improved by employing more efficient optimization methods 528 than the adopted trial and error method. Furthermore, the 529 scheduling rules can be extended by incorporating the angular 530 parameter from the radar echo to enhance the performance of 531 the scheduler.  He is a Distinguished Lecturer and a member of the Turkish Academy of 759 Sciences. In addition, he has worked as a part-time Consultant for various 760 companies and institutions, including Anritsu Company and The Scientific 761 and Technological Research Council of Turkey. He conducts research in 762 wireless systems, with emphasis on the physical and medium access layers 763 of communications. His current research interests include 6G and beyond 764 radio access technologies, physical layer security, interference management 765 (avoidance, awareness, and cancelation), cognitive radio, multi-carrier wire-766 less technologies (beyond OFDM), dynamic spectrum access, co-existence 767 issues, non-terrestrial communications (high altitude platforms), joint radar 768 (sensing), and communication designs. He has been collaborating exten-769 sively with key national and international industrial partners and his research 770 has generated significant interest in companies, such as InterDigital, Anritsu, 771 NTT DoCoMo, Raytheon, Honeywell, and Keysight Technologies. Collab-772 orations and feedback from industry partners has significantly influenced 773 his research. In addition to his research activities, he has also contributed 774 to wireless communication education. He has integrated the outcomes of 775 his research into education which lead him to develop a number of courses 776 at the University of South Florida and Istanbul Medipol University. He has 777 developed a unique ''Wireless Systems Laboratory'' course (funded by the 778 National Science Foundation and Keysight Technologies) where he was 779 able to teach not only the theory but also the practical aspects of wireless 780 communication system with the most contemporary test and measurement 781 equipment.

782
Dr. Arslan has served as the general chair, the technical program com-783 mittee chair, a session and symposium organizer, the workshop chair, and a 784 technical program committee member for several IEEE conferences. He is 785 currently a member of the Editorial Board for the IEEE COMMUNICATIONS 786 SURVEYS AND TUTORIALS and the Sensors journal. He has also served as a mem-787 ber of the Editorial Board for the IEEE TRANSACTIONS ON