同济大学：《传播信道特征估计和建模》课程教学资源（教案讲义）Chapter 09 Practices - channel modeling for modern communication systems.pdf

9 Practices: channel modeling for modern communication systems Based on the knowledge introduced in previous chapters, this chapter will present the development of channel models for various communication scenarios, e.g., conventional cellular scenarios, newly emerging V2V scenarios and cooperative scenarios, via different modeling approaches. From this chapter, some new channel models will be demonstrated and some interest observations and conclusions will be given. 9.1 Typical scenarios for vehicle-to-vehicle and cooperative communications ((10.1)) In general, different communication scenarios result in different channel characteristics. Therefore, to investigate underlying channel characteristics of any communication systems, it is desirable to completely classify the corresponding typical scenarios, on which the communication systems are developed. Unlike newly emerging communication systems, e.g., V2V communications and cooperative communications systems, typical scenarios for conventional cellular communication systems have been completely classified and widely accepted in public. In general, typical scenarios for conventional cellular communications can be classified as Macro-cell scenarios, Microcell scenarios, and Pico-cell Scenarios. As newly emerging communication systems, typical scenarios for V2V and cooperative communications are different compared with those for conventional cellular systems, resulting in unique channel characteristics of these newly emerging communication systems. This section will introduce and classify typical scenarios for the newly emerging V2V and cooperative communication systems. 9.1.1 Vehicle-to-vehicle communication scenarios ((10.1.1)) Knowledge of the V2V propagation channel for different scenarios is of great importance for the design and analysis of V2V systems. However, due to the large difference between the conventional cellular and V2V channels, the channel knowledge obtained from conventional cellular systems cannot be directly used for V2V systems. So far, some measurement campaigns have been conducted and others are ongoing to investigate the V2V propagation channels for different application scenarios. Here, we will briefly review and classify some recent typical V2V scenarios according to carrier frequencies, frequency-selectivity, antennas, environments, Tx/Rx direction of motion, and channel statistics, as shown in Table 10.1. • Carrier Frequencies Before the IEEE 802.11p standard ? was proposed, some measurement campaigns were conducted at carrier frequencies outside the 5.9 GHz dedicated short range communication (DSRC) band. In Acosta et al. (2004); ?, V2V measurements were carried out at 2.4 GHz, i.e., the IEEE 802.11b/g band. Some measurements were done around the IEEE 802.11a frequency band, e.g., at 5 GHz in ? and at 5.2 GHz in ?. Measurements at 5.9 GHz were presented in ?? for narrowband and wideband V2V channels, respectively. Based on the aforementioned measurement campaigns, we can observe that propagation phenomenon in similar environment with different frequencies can vary significantly. Therefore, more measurement campaigns are expected to be conducted at 5.9 GHz for the better design of safety applications for V2V systems following the IEEE 802.11p standard. On the other hand, for improved design of nonsafety applications for V2V systems, measurement campaigns performed at other frequency bands, e.g., 2.4 GHz or 5.2 GHz, are still required. • Frequency-Selectivity and Antennas This is a Book Title Name of the Author/Editor c XXXX John Wiley & Sons, Ltd

256 Practices: channel modeling for modern communication systems Table 10.1. Important V2V channel measurements. SS: suburban street; EW: expressway; UC: urban canyon; Micro: Micro-cell; Pico: Pico-cell; H(L)VTD: high (low) vehicular traffic density; PDP: power delay profile; DD: Doppler-delay; PSD: power spectrum density; STF: space-time-frequency; CF: correlation function; LCR: level crossing rate; SDF: space-Doppler-frequency; PDF: probability density function; PL: path loss; CDF: cumulative distribution function; CT: coherence time. In 1999, the Federal Communications Commission (FCC) allocated 75 MHz of licensed spectrum, including seven channels, each with approximately 10 MHz instantaneous bandwidth, for DSRC in the USA. Such V2V channels are often frequency-selective (wideband) channels. A narrowband fading channel characterisation based on measurement results, e.g., in ?, is not sufficient for such V2V DSRC applications. Wideband measurement campaigns, e.g., in Acosta et al. (2004); ?); ?); ?); ?, are therefore essential for understanding the frequency-selectivity features of V2V channels and further designing high-performance V2V systems. Most V2V measurement campaigns so far have focused on single-antenna applications, resulting in SISO systems, e.g., in Acosta et al. (2004); ?); ?); ?. MIMO systems, with multiple antennas at both ends, are very promising candidates for future communication systems and are gaining more importance in IEEE 802.11 standards. Moreover, MIMO technology becomes more attractive for V2V systems since multiple antenna elements can be easily placed on large vehicle surfaces. However, until now only a few measurement campaigns, e.g., in ??, were conducted for MIMO V2V channels. Hence, more MIMO V2V wideband measurement campaigns are needed for future V2V system developments. • Environments and Tx/Rx Direction of Motion Similar to conventional cellular systems, V2V scenarios can be classified as large spatial scale (LSS), moderate spatial scale (MSS), and small spatial scale (SSS) according to the Tx–Rx distance. For LSS scenarios or MSS scenarios, where the Tx–Rx distance is normally larger than 1 km or ranges from 300 meters to 1 km, V2V systems are mainly used for broadcasting or geocasting, i.e., geographic broadcasting ?. While for SSS scenarios, where the Tx–Rx distance is usually smaller than 300 meters, V2V systems can be applied to broadcasting, geocasting, or unicasting. Since most V2V applications fall into MSS or SSS scenarios, these two scenarios are currently receiving more and more attention with several current measurement campaigns taking place, e.g., in Acosta et al. (2004)–?. However, there are still few applications that need communications between two vehicles separated by large distances, e.g., larger than 1 km. For such LSS V2V applications, one example is V2V decentralized environmental notification, which means that vehicles or drivers in a certain area share information with each other about observed events or roadway features. These applications have not gained much attention and thus no measurement results are available that explore V2V channels for LSS scenarios. V2V scenarios can also be categorized as urban canyon, suburban street, and expressway in terms of roadside environments, i.e., buildings, bridges, trees, parked cars, etc., located on the roadside. Many measurement campaigns, e.g., in Acosta et al. (2004)–? and ?, were conducted to study the channel statistics for various types of roadside environments. Due to the unique feature of V2V environments, the vehicular traffic density (VTD) also significantly affects the channel statistics, especially for MSS and SSS scenarios. In general, the smaller Tx–Rx distance, the larger impact of the VTD. Note that V2V channels usually exhibit non-isotropic scattering except in cases of high VTD. To the best of the authors’ knowledge, only one measurement campaign ? was carried out to study the impact of the VTD for expressway MSS and SSS scenarios. Directions of motion of the Tx and Rx also affect channel statistics, e.g., Doppler effects. Many measurement campaigns, e.g., in Acosta et al. (2004)–? and ?, have focused on studying channel characteristics when the Tx and Rx are moving in the same direction. Few V2V measurement campaigns, e.g., in ? and ?, have investigated channel characteristics when the Tx and Rx are moving in opposite directions. In summary, it is desirable to conduct more measurement campaigns for MSS and SSS scenarios with various VTDs when the Tx and Rx move in opposite directions. In addition, measurement campaigns for LSS scenarios are indispensable for some V2V applications that need communications between two vehicles with a large distance. 9.1.2 Cooperative communication scenarios ((10.1.2)) Cooperative MIMO groups multiple radio devices to form virtual antenna arrays so that they can cooperate with each other by exploiting the spatial domain of mobile fading channels. Note that the cooperation of the grouped devices does not mean that the base station (BS) and mobile station (MS) have to be in reach of each other. Some devices

Practices: channel modeling for modern communication systems 257 can be treated as relays to help the communication between the BS and MS. Unlike conventional point-to-point MIMO systems, cooperative MIMO systems consist of multiple radio links, e.g., BS-BS, BS-relay station (RS), RS-RS, RS-MS, BS-MS, and MS-MS links. Due to different local scattering environments around BSs, RSs, and MSs, a high degree of link heterogeneity or variations is expected in cooperative MIMO systems. In this chapter, we are interested in various cooperative MIMO environments which can be classified based on the physical scenarios and application scenarios. The physical scenarios include outdoor macro-cell, micro-cell, pico-cell, and indoor scenarios. Each physical scenario further includes 3 application scenarios, i.e., BS cooperation, MS cooperation, and relay cooperation. Therefore, 12 cooperative MIMO scenarios are considered in this chapter. 9.2 Channel characteristics ((10.2)) Knowledge of channel statistics is essential for the analysis and design of a communication system. Unlike a rich and fascinating history of the research of conventional cellular channel characteristics, the investigation of channel characteristics for emerging V2V and cooperative MIMO systems is still in its infancy. This section will introduce the important channel characteristics of V2V and cooperative communication systems. 9.2.1 Channel characteristics of V2V communication systems ((10.2.1)) As shown in Table 10.1, many different V2V channel statistics have been studied in recent measurement campaigns Acosta et al. (2004)–?. Here, we only concentrate on two important statistics, amplitude distribution and Doppler power spectral density (PSD). Analysis of amplitude distributions has been reported in ?, ?, and ?. In ?, the authors modeled the amplitude probability density function (PDF) of the received signal as either Rayleigh or Ricean. In ?, it was observed that the received amplitude distribution in a dedicated V2V system with a carrier frequency of 5.9 GHz gradually transits from near-Ricean to Rayleigh as the vehicle separation increases. When the line-of-sight (LoS) component is intermittently lost at large distances, the channel fading can become more severe than Rayleigh. A similar conclusion has been drawn in ?, where the amplitude PDF is modeled as Weibull distribution and this “worse than Rayleigh” fading is called severe fading. The reason behind the severe fading is the rapid transitions of multipath components induced by high speed and low height of the Tx/Rx and fast moving scatterers ?. The Doppler PSD has been investigated in Acosta et al. (2004), ?, and ?–?. Joint Doppler-delay PSD measurements for wideband V2V channels at 2.4 GHz, 5.2 GHz, and 5.9 GHz were reported in Acosta et al. (2004), ?, and ?, respectively. It was demonstrated that Doppler PSDs can vary significantly with different time delays in a wideband V2V channel. In ?, the authors analyzed the Doppler spread and coherence time of narrowband V2V channels and presented their dependence on both velocity and vehicle separation. Recently, the space-Doppler PSD, which is the Fourier transform of the space-time correlation function in terms of time, was investigated in ? and ?. It is worth noting that the Doppler PSD for V2V channels can be significantly different from the traditional U-shaped Doppler PSD for F2M channels. 9.2.2 Channel characteristics of cooperative communication systems ((10.2.2)) Several papers have reported measurements of various statistical properties of cooperative MIMO channels for different scenarios. A few indoor cooperative channel measurements were reported in ??, where the cooperative nodes are all static. Mobile multi-link measurements were presented in ? for indoor cooperative MIMO channels. Outdoor cooperative MIMO channel measurements were addressed in ?? and ?–? for static nodes and mobile nodes, respectively. All these measurement campaigns concentrated on the investigation of the channel characteristics of individual links for different scenarios, such as path loss, shadow fading, and small scale fading. As mentioned before, cooperative MIMO systems include multiple radio links that may exhibit strong correlations. The correlation of multiple links exists due to the environment similarity arising from common shadowing objects and scatterers contributing to different links and can significantly affect the performance of cooperative MIMO systems. The investigation of the correlations between different links is rare in the current literature and thus deserves more investigation. 9.3 Scattering theoretical channel models for conventional cellular MIMO systems ((10.3)) The well-known one-ring MIMO F2M RS-GBSM ? has been widely used for the analysis and design of narrowband MIMO cellular systems for macro-cell scenarios. To reach the high demand for high-speed communications, wideband

258 Practices: channel modeling for modern communication systems MIMO cellular systems have been suggested in many communication standards, leading to the increasing requirement for wideband MIMO F2M channel models. However, the one-ring structure that assumes effective scatterers located on a one-ring is overly simplistic and thus unrealistic for modeling wideband channels ?. Therefore, in ? the authors for the first time extended the narrowband one-ring model to a wideband model by extending the location of effective scatterers on a one-ring to a circularring. However, it is not trivial to use this model to match any given or measured PDP since many parameters need adjustments via a complicated approach. In addition, the integral expressions of the derived STF CFs based on this model can only be numerically evaluated as no closed-form expressions were found. In contrast to ?, in ? the one-ring model was extended to a wideband application by dividing the one-ring into several segments in terms of different time of arrivals (ToAs). This makes the model easier to match any specific PDPs due to the TDL structure. However, the onering structure was still applied in this model, which causes this model to exhibit an unrealistic structure in that certain ToAs (or propagation delays) are always related to a certain proportion of AoAs. To obtain closed-form expressions of STF CFs, the model in ? further applied an unrealistic assumption that the mean AoA and the corresponding angle spread are exactly the same for all the scatterers in different segments. Moreover, neither of the two models considered the interaction of the AoA, AoD, and ToA, the importance of which was described in ?. More importantly, frequency correlation of sub-channels with different carrier frequencies, studied in ? for SISO channels, has not been investigated yet for MIMO channels. Chapter 3 showed that the frequency correlation appears when the coherence bandwidth Bc is larger than the frequency separation of different sub-channels. This frequency-correlated MIMO channel is commonly encountered in frequency-diversity MIMO communication systems, such as frequency-hopping MIMO systems and MIMO-orthogonal frequency division multiplexing (MIMO-OFDM) systems. For convenience, researchers investigating these systems have typically assumed that different frequency-separated channels are uncorrelated ?, which leads to positively biased performance results for the investigated systems. Therefore, accurate theoretical analysis and simulations of frequency-correlated MIMO channels are of great importance in investigating the impact of the frequency correlation on the performance of real frequency-diversity MIMO systems. The goals of this section are three-fold. First, we propose a new wideband MIMO F2M RS-GBSM that represents a reasonable compromise between physical reality and analytical tractability. The proposed model uses a concentric multiple-ring instead of one-ring around the MS to avoid the one-ring structure. Also, to easily match any specified or measured PDP, the model utilizes a virtual confocal multiple-ellipse to construct a TDL structure. Moreover, this model has the ability to consider the interaction of the AoA, AoD, and ToA. Secondly, from the proposed model, a closed-form expression of the STF CF between any two sub-channels with different carrier frequencies for each timebin signal is derived. For simplicity, the spatial and frequency correlations were assumed to be independent in ??. On the contrary, in this subsection we derive the STF CF by taking into account the dependency between them, which is a unique characteristic of frequency-correlated MIMO channels ?. Therefore, the derived STF CF can explicitly relate the frequency correlation to various environment parameters (e.g., mean AoA and angle spread of the AoA). Moreover, based on the derived CF, we reveal the inherent frequency correlation within the spatial correlation, which is important for the design of frequency-diversity MIMO systems ??. In addition, the proposed wideband model can be reduced to the traditional one-ring model by removing the frequency-selectivity. In this case, the derived STF CF shows a compact closed-form expression that is a generalisation of many existing CFs ??????. Since the proposed wideband RS-GBSM is a reference model that assumes an infinite number of effective scatterers, it cannot be implemented directly in practice. Therefore, the third goal of this section is to derive a wideband deterministic SoS simulation model based on the proposed reference model. Closed-form expressions are provided for the STF CF of the simulation model. Similarly, the corresponding narrowband simulation model can be obtained by removing the frequency-selectivity from the wideband simulation model. The statistical properties of our simulation models are verified by comparing with the corresponding statistical properties of the reference models. 9.3.1 A Wideband Multiple-ring Based MIMO Channel Reference Model ((10.3.1)) A narrowband one-ring MIMO model is suitable for describing a narrowband channel in macro-cell scenarios, where the BS is elevated and unobstructed, while the MS is surrounded by a large number of local scatterers. For narrowband systems, since the propagation delays τ ′ n of all N (N→∞) incoming waves are much smaller than the transmitted symbol duration Ts, i.e., τ ′ max=max {τ ′ n} N n=1≪Ts, the delay differences caused by different local scatterers randomly around the MS can be neglected in comparison to Ts. Therefore, it is reasonable to use the effective scatterers located on a one-ring instead of the real local scatterers to construct the low complexity one-ring model at the minor expense of accuracy ?. However, in high data rate wideband systems, Ts is much smaller than that in narrowband systems. In this case, the propagation delay differences cannot be neglected and thus the channel becomes a wideband channel

Practices: channel modeling for modern communication systems 259 l τ ′ o T ′ Θ ,il R δ δ T R ,,, nril φ T ,, nil φ β T R ,il q R ′ l i on , , ε l i nq , , ε Y v l i r , , µ l i, ∆µ −1 ′ l τ +1 ′ l τ D X ,il +1 R β R R il −1, l i n , , ξ γ o T il S , nil S ,, q R Fig. 9.1 A new wideband multiple-ring based MIMO F2M channel model. Therefore, the one-ring structure violates the basic characteristics of wideband channels as mentioned in ?. To extend the one-ring model to wideband applications in macro-cell scenarios, the primary task is to modify the oversimple one-ring structure. To this end, we replace the one ring of effective scatterers by concentric multiple rings of effective scatterers around the MS to capture the basic characteristics of wideband channels. To make our model easier to match any specified or measured PDP, we utilise the confocal multiple virtual ellipses with the BS and MS located at the foci to represent the TDL structure, where different delays correspond to different virtual confocal ellipses (i.e., taps). Note that the total number of virtual confocal ellipses L and the values of major axes a of different ellipses are determined according to the specified or measured PDP. The newly developed structure is shown in FIGURE 10.2. For clarity, FIGURE 10.2 only presents the effective scatterers from three concentric rings (the total number of concentric rings in the lth tap is Λl) belong to the lth tap (l=1, 2, ..., L). Notice that the total number and radii of the concentric multiple rings for different taps (i.e., Λl and Rl,i) can be different in terms of different propagation environments for different delays. Therefore, the new wideband model with the appropriate number and major axes of virtual confocal multiple ellipses (i.e., L and a), and the appropriate number and radii of concentric multiple rings (i.e., Λl and Rl,i) should be suitable for any macro-cell scenario. We assume that uniform linear antenna arrays are used with MT=MR=2 antennas. The symbols δT and δR designate the antenna element spacing at the BS P and MS, respectively, and D denotes the distance between the BS and MS. The effective scatterers are located on L l=1 Λl rings with radii Rl,i (i=0, 1, ...,Λl − 1). It is usually assumed that the assumption D≫Rl,i≫max {δT , δR} is fulfilled. The multi-element antenna tilt angles are denoted by βT and βR. The MS moves with speed in the direction determined by the angle of motion γ. The angle spread seen at the BS is denoted by Θl.i, which is related to Rl,i and D by Θl,i ≈ arctan(Rl,i/D) ≈ Rl,i/D. The received complex impulse response at the carrier frequency fc for the link To − Rq can be expressed as hoq (t, τ′ ) = X L l=1 hl,oq (t) ×δ (τ ′ − τ ′ l ) (9.1) with o = 1, 2, ..., MT and q = 1, 2, ..., MR, where hl,oq(t) and τ ′ l denote the complex time-variant tap coefficient and the