Communication Requirements for Cooperative Adaptive Cruise Control

Transcription

1 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 1 Communication Requirements for Cooperative Adaptive Cruise Control Michał Sybis, Member, IEEE, Vladimir Vukadinovic, Marcin Rodziewicz, Paweł Sroka, Member, IEEE, Adrian Langowski, Karolina Lenarska, and Krzysztof Wesołowski, Member, IEEE Abstract The paper investigates the impact of vehicle-tovehicle communications on the ability of cooperative adaptive cruise control to support high-density car platooning. We first use a simplistic communication model to study the impact of actuation lag, message periodicity, and communication delay on the minimum feasible inter-car spacing. We then use a detailed IEEE p simulation model to evaluate the platooning performance in realistic highway scenarios. Different highway traffic intensities are simulated to observe the impact of increasing contention on the wireless channel with two different transceiver configurations: with a single-transceiver operating on the common safety channel and with a dual-transceiver operating simultaniously on the common safety channel and a dedicated service channel. Index Terms Cruise Control, Adaptive Cruise Control, Cooperative Adaptive Cruise Control, Communication Requirements I. INTRODUCTION VEHICLE platooning is coordinated movement of a group of vehicles forming a convoy led by a platoon leader. It was conceived as a way to increase road capacity, reduce fuel consumption, and improve driver safety and comfort. Studies have shown that platooning may double the road capacity [1]. To achieve this, platooning shall allow close spacing between vehicles while they move at highway speeds. The limiting factor in this case is driver s capability to safely maintain the close spacing. Automated drive control supported by onboard sensors and inter-vehicular communications can provide much faster and more precise response than human drivers. Therefore, platooning relies on drive automation to eliminate human errors and improve road safety. Platooning has the potential to reduce fuel consumption and exhaust emissions due to the reduced air drag. Road freight transport using trucks would benefit greatly since fuel costs account on average for 25-30% of a truck s operational costs [2]. The findings of the SARTRE project [3] show that platooning provides fuel savings from 7 to 15 % for trucks travelling behind the leader. The fuel savings translate to substantial reduction of CO 2 emission. From an implementation point of view, platooning is very challenging. It requires an interplay of drive control algo- M. Sybis, M. Rodziewicz, P. Sroka, A. Langowski, K. Lenarska and K. Wesolowski are with the Faculty of Electronics and Telecommunications of Poznan University of Technology, Poznan, Poland, e- mail:{michal.sybis, Marcin.Rodziewicz, Pawel.Sroka, Adrian.Langowski, Karolina.Lenarska, Krzysztof.Wesolowski}@put.poznan.pl V. Vukadinovic is with Nokia, Wroclaw, Poland, {Vladimir.Vukadinovic}@nokia.com. rithms, actuation, sensing, and wireless communication. The first important step was taken twenty years ago with the introduction of Adaptive Cruise Control (ACC). ACC adapts the acceleration of the host vehicle to maintain a desired headway time to the preceding vehicle, using information supplied by on-board sensors (e.g. radar, lidar, and camera). ACC is an autonomous controller: it does not depend on inter-vehicular communication or any form of cooperation among vehicles. However, ACC is not able to provide stable and robust performance with close spacing between vehicles irrespective of their speed, as discussed later in the paper. Therefore, research community started to work on an extension of ACC called Cooperative Adaptive Cruise Control (CACC), which relies both on sensors measurements and wireless communication with other vehicles in the platoon. Depending on the communications topology, different realizations of CACC are possible. Typically, a platoon leader broadcasts its speed and acceleration to all vehicles in the platoon, allowing them to react faster to any changes in traffic conditions compared to relying on on-board sensors only. Several examples of CACC designs with different communication topologies and control policies are described in [4]-[7]. A comprehensive survey is provided in [8]. All these algorithms require frequent, highlyreliable, and low-latency Vehicle-to-Vehicle (V2V) communication. In this paper, we first investigate the communications requirements of the sliding-surface CACC controller [9] in terms of message frequency and latency using an abstract communication model. The case of ideal (zero-loss and zero-latency) communication is a benchmark for evaluating the impact of communications impairments on minimum feasible intercar spacing. The interplay between communication latency and actuation lag is analyzed and a method to eliminate the actuation lag from the CACC control loop is proposed in order to improve platoon stability and minimize inter-car spacing. We then study the ability of the existing communication technologies to support CACC. While it is expected that future road safety and traffic efficiency applications will rely on a combination of different communication technologies (802.11p, LTE, and 5G), the two currently existing sets of standards for ITS, namely DSRC in U.S. [10] and C-ITS in Europe [11], both assume IEEE p at their PHY and MAC layers. Therefore, in this paper, we focus on p standard and its extension for multi-channel operations [12]. Using detailed simulations, we analyze the performance of a platoon of ten cars travelling on a highway. In the con- c 2017 IEEE with the single p transceivers, the input /00$00.00figuration to

2 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 2 the CACC controller is provided via Basic Safety Messages (BSMs) that are broadcasted by all vehicles on a channel dedicated to traffic safety messages (CH172). We consider different highway traffic densities and their impact on the reliability of BSM reception. In the configuration with two p transceiver, where the second transceiver is tuned to a dedicated service channel (SCH), the input to the CACC controler is provided both by BSMs broadcasted on CH172 and by special CACC messages broadcasted by platooned cars on the SCH channel, where no other transmissions take place. Our results show that the single transceiver configuration is not able to provide reliable platooning with short inter-car spacing when traffic densities on the highway is moderate-to-high. The paper is organized as follows: An overview of related work is provided in Section II. Theory behind ACC and CACC is summarized in Section III. Section IV introduces the model of the platoon system under evaluation. The study of communication requirements of CACC and the performance evaluation of p-enabled platooning are provided respectively in Sections V and VI. Section VII concludes the paper. II. RELATED WORK A number of empirical studies have been performed to evaluate and demonstrate the performance of car/truck platooning based on p/DSRC communication. The SARTRE project [3] deployed a platoon of two trucks and three cars driven autonomously in close formation. The experiments showed that the platoon can drive at speeds of up to 90 km/h with a 5-7 m gap between the vehicles. A platoon of three fully-automated trucks with 10 m inter-vehicle distance was tested on an expressway in Japan in the Energy ITS project [13]. European Truck Platooning Challenge 2016 [14] was the first successful experiment with cross-border platooning: Automated trucks of six major truck vendors have been driving in platoons on public roads from several European cities to Rotterdam in the Netherlands. While most vendors did not publish details as to the underlying V2V equipment, at least DAF, MAN, and Daimler trucks were equipped with pbased technologies. The PATH Program of UC Berkeley and Volvo demonstrated a platoon of three DSRC-equipped trucks driving 15 meters apart on busy 110 Interstate freeway in 2017 [15]. While trials certainly provide valuable insight into platooning performance under realistic radio propagation conditions and with real-life vehicle dynamics, they do not account for a scenario where all other vehicles on the highway will be equipped with DSRC and contending for the same radio resources used for intra-platoon communication. Therefore, simulations are still an indispensable tool to study how DSRC performance degrades with highway traffic density and how inter-vehicle spacing in the platoon shall be adapted to account for this degradation. As an example of this research direction, [16] builds a comprehensive simulation framework to study the CACC performance in the presence of imperfect communication. The study shows that the (BSM/CAM) message broadcast frequency and loss ratio have significant influence on the string stability of the evaluated CACC controller. The authors in [17] compare different message broadcasting/beaconing solutions for platooning i.e. Dynamic Beaconing (DynB) and ETSI Decentralized Congestion Control (DCC) and propose their own solution that couples contention-free TDMA-like approach with transmission power control. The proposed solution eliminates the channel contention by allowing platooned vehicles to negotiate a TDMA schedule on top of standard p channel access method. A similar approach is explored in [18], where TDMA is realized using a token passing among platoon members. The benefit of TDMA is maximized when intra-platoon communication takes place on a dedicated service channel. However, when platooning relies on BSMs/CAMs transmitted on a common control/safety channel used also by surrounding vehicles, which use standard contention-based p channel access method rather than TDMA, its value diminishes. A comparison of platooning performance with intra-platoon communication on a common control/safety channel and on a dedicated service channel is provided in [19]. Communication on a dedicated service channel provides superior performance due to reduced contention and less stringent requirements on sending rates, message types/sizes, and medium access methods. In this paper, we show how dual transceiver configuration can be used to further improve the platooning performance through multi-channel operations i.e. by exploiting both the BSMs transmitted on the common safety channel and special CACC messages transmitted on a dedicated service channel. Another approach for p multi-channel operation, which assumes singletransceiver configuration, is presented in [20]. The approach utilizes information about channel loads and service priorities to control the timing of channel switching between the control/safety channel and service channel. We do not consider multi-channel operation (i.e. channel switching) with a single transceiver due to the assumption that the single transceiver must be continuously tuned to control/safety channel to receive BSMs of surrounding vehicles. The impact of kinematic triggering of BSM/CAM generation on platooning performance is investigated in [21]. Triggering conditions are based on the dynamics of an originating vehicle: current speed, position, and direction are tracked and compared to the values sent in the last triggered CAM. If the differences exceeded any of the pre-defined thresholds, a new CAM/BSM is generated. Since kinematic triggering may lead to negative CAM/BSM synchronization effects that increases the contention for wireless channel access, we assume that BSMs are generated at regular intervals. III. OVERVIEW OF CRUISE CONTROL ALGORITHMS This section introduces the theory behind ACC and CACC. A. Adaptive Cruise Control ACC controls the vehicle speed to match the value set by the driver when no lead vehicle is in sight. When a slower leading vehicle is present, the ACC controlled vehicle will follow the lead vehicle at the safe distance. In this way, vehicles using the same lane of a highway could form a platoon. Research on ACC began in the 1960s. It has been shown [9]

3 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 3 Fig. 1. An example of a part of a car platoon Fig. 3. ACC system with constant desired headway time between cars a des is delayed and can be well approximated by the first order inertial system, so the plant transfer function is Fig. 2. ACC system with constant desired distance between cars that Proportional Integral Derivative (PID) controllers used to set the command for acceleration or deceleration or their variations produce satisfactory control results in ACC systems. Let us consider the platoon of cars shown in Fig. 1. There are two specifications that the vehicle following control system has to satisfy: individual vehicle stability and string stability [9]. The vehicle controlled by ACC is individually stable if its spacing error converges to zero when the preceding vehicle is operating at a constant speed. The spacing error is the difference between the actual spacing from the preceding vehicle and the desired inter-vehicle spacing L des. If the preceding vehicle is accelerating or decelerating then the spacing error is non-zero. The spacing error, δ i, for the i-th vehicle in the platoon can be described by the formula δ i = x i x i 1 +L des. (1) where x i is the position of the i-th vehicle in the platoon. The desired spacing, L des, includes the preceding vehicle length, l i 1. The ACC control provides individual vehicle stability if the following condition is fulfilled if a i 1 = ẍ x 1 0 then δ i 0. (2) Now in the case of string stability, spacing errors are guaranteed not to amplify as they propagate towards the end of the string. In typical ACC systems, the control is performed in two levels. The upper level controller determines the desired acceleration, a des (t) = ẍ des, for each vehicle in such a way that the two above mentioned specifications (individual vehicle stability and string stability) are met. The lower layer controller determines the throttle and brake commands needed to track the desired acceleration. Let us consider two particular cases of the control system for ACC. The first one is shown in Fig. 2. In this case the aim of the controller is to maintain the desired constant value of inter-vehicle spacing L des. The error signal used to control the controller is then described by (1). Let us assume that acceleration of the controlled vehicle can be instantaneously controlled, i.e. a des,i = ẍ des,i = a i = ẍ i This means that the plant (lower layer controller) transfer function is equal to one (P(s) = 1). This is a simplification of reality. Typically, the response of the car a i to the command P(s) = 1 τs+1. (3) Let δ i and δ i 1 be the spacing errors of consecutive ACC controlled vehicles in a string (platoon). We recall that a string remains stable if spacing errors do not amplify along the string from its beginning to the end. LetĤ(s) be the transfer function related to spacing errors of consecutive vehicles, i.e. Ĥ(s) = i(s) i 1 (s), where i(s) = L[δ i (t)]. (4) The string is stable if the following conditions are fulfilled [9]: Ĥ(s) 1, The impulse response corresponding to Ĥ(s) does not 1 change its sign, i.e. L [Ĥ(s)] = h(t) > 0, for t 0. For string stability in the case when the constant spacing between vehicles is to be maintained and the car responds to the command ideally, i.e. P(s) = 1, one can prove that the constant spacing policy in ACC is not suitable for use if each vehicle is controlled individually by the ACC system. The precise proof of this statement can be found in [9]. For the above reason another policy has been selected which ensures individual vehicle stability and string stability at the same time. This policy is Constant Time-Gap (CTG) policy (Fig. 3), also known as constant headway time policy. In this policy the desired inter-vehicle spacing depends linearly on vehicle velocity. Thus, the spacing error is given by the formula δ i = x i x i 1 +l i 1 +hẋ i = ε i +hẋ i. (5) where ε i = x i x i 1. The following controller equation implementing the CTG rule was developed by Ioannu and Chien [22]: ẍ des,i = 1 h ( ε i +λδ i ). (6) Parameter h has a meaning of the headway time. It is time needed by the i-th vehicle to reach the preceding one. Let us assume that the lower layer controller reacts like a first order inertial system. Thus, its dynamics are described by the differential equation of the form: τ... x i +ẍ i = ẍ des,i. (7) Substituting the desired acceleration from equation (6) in (7) we obtain τ... x i +ẍ i = 1 h ( ε i +λδ i ). (8)

4 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 4 Calculating the second derivative of (5) we get δ i = ε i +h... x i, (9) so taking... x i from (8) and using it in (9) we obtain ε i = δ i + 1 τ (ε i +hẍ i +λδ i ), (10) or equivalently: ε i = δ i + 1 ) ( δi +λδ i. (11) τ The difference between distance errors of successive vehicles is: δ i δ i 1 = ε i ε i 1 +h(ẋ i ẋ i 1 ) = ε i ε i 1 +h ε i. (12) Calculating Laplace transform of (11) and (12) and using E i (s) = L[ε i ] and E i 1 (s) = L[ε i 1 ] in equation (12) represented in Laplace domain we obtain the ratio: L[δ i ] L[δ i 1 ] = i(s) i 1 (s) = s+λ hτs 3 +hs 2 +(1+λh)s+λ. (13) It turns out that transfer function (13) characterizing the string stability when substituting s = jω and calculating its magnitude, is smaller than one for any frequency if h 2τ. (14) Therefore, the string using the constant time gap (headway) policy for which the time gap meets (14) remains stable. As we see in (14), the inertial (first order time) constant τ plays a crucial role in the dynamics of the string of cars and determines their density in the platoon if ACC control mode is applied. In [9] one can find a discussion on a typical value of τ which can be expected in reality. So far we have not mentioned any hard limits for the minimum or maximum distance, and in consequence, distance errors between cars. Setting such limits causes the control process to be nonlinear and its analysis to be much more difficult. To conclude, it seems that the further development of the ACC algorithms to ensure individual vehicle and string stability will rely on the application of sophisticated nonlinear algorithms going beyond the linear controllers discussed in this section. Publications [23] and [24] are examples of such approaches. Nevertheless, in our study the algorithm described by (6) and (7) is applied when a car operates in the ACC mode. B. Cooperative Adaptive Cruise Control Cooperative adaptive cruise control is a natural extension of adaptive cruise control based on additional information obtained via a wireless link from vehicles travelling in the same platoon. Although the topic of CACC is relatively new, it attracts the attention of many academic and industrial research centers. Several papers can be found in the literature (see [25], [26], [27], [28] as the examples), including PhD dissertations such as [29]. In [29] the author considers CACC controllers that, in addition to sensor measurements, receive data from one or two preceding cars wirelessly. In a majority of papers, the Fig. 4. A heterogeneous platoon with communication from preceding vehicles additional data is the acceleration of the preceding car only. In more complex scenarios, the leading vehicle transmits its data to all other vehicles in the platoon. In reality, each car in the platoon can announce its position, acceleration and other parameters to all other cars within the platoon. A platoon can consist of identical vehicles, however, more probable is the situation in which different types of vehicles constitute a platoon. In the first case we talk about homogeneous platoon, whereas in the latter case the platoon is called heterogeneous (see Fig. (4)). Similarly as for the ACC mode, in the CACC mode several spacing policies can be considered. The first one is constant spacing policy, however, careful design of the controller parameters has to be employed to ensure the string of vehicles to be stable. The popular policy is constant time spacing policy (constant headway time considered previously). Using this rule we can ensure string stability, as in the ACC case. The third policy is a nonlinear spacing policy. The first report of which was published in 1998 [30]. However, it is still a subject of research [25]. The CACC system often consists of feedback and feedforward blocks. In literature several types of feedback controllers in CACC systems can be found. The most popular is a PD controller [29]. As declared in [25] and [29], when the PD feedback controller is applied, string stability is achieved by a posteriori tuning the controller parameters. Other types of controllers are Model Predictive Controller [31] and Linear- Quadratic Regulator [32]. Feedforward controllers are applied to minimize the distance error [29]. They can also be used to ensure string stability. Assuming a heterogeneous vehicle platoon, the controller designer has to be aware that each car participating in the platoon can have different parameters such as communication delay time, vehicle lag, actuator delay time and other constraints. Depending on the assumed spacing policy, an appropriate analysis of string stability can be performed. In case of headway time policy, such analysis can be found in [25] and [29]. In [25] the authors conclude that string stability of a heterogeneous platoon is influenced by the vehicle time constant and communication delay time. Choosing different controller parameters and headway time for each vehicle in such platoons can ensure string stability. In [25] another intuitive conclusion is made that controller parameters and the minimum headway time of vehicles with larger communication delay time should be chosen larger than those of vehicles with low communication delay. In general, the value of the system latency (not only due to coding and signal processing but also due to the delay introduced by multiple access contention or contention free method applied in the

5 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 5 communication system) can have significant influence on both stability of the platoon and its physical length (determined by selection of necessary headway times). Detailed analysis of CACC shall take into account unreliable wireless communication. Such analysis has been reported in [26]. If the wireless inter-vehicle communication fails, CACC would automatically degrade to ACC, leading to a significant increase in minimal headway time to maintain string stability. In [26] it is proposed to estimate the actual acceleration of the preceding vehicle that could replace the transmitted value of this acceleration. String stability in such a case is also considered. Let us consider constant spacing policy supported by wireless communication within a platoon. This policy is particularly attractive as it leads to the highest road traffic efficiency and fuel savings. As already mentioned, if a CACC algorithm applied in the i-th car in the platoon relies on its sensors and information received, e.g. via a wireless link from the preceding car only, we have an autonomous controller in each vehicle and the headway time has to be assumed as the criterion to set the CACC controller parameters in order to enable string stability. However, if the i-th vehicle receives information from more cars than the preceding vehicle only, constant spacing policy can be applied [9, Section 7.7]. For that purpose the so called sliding surface controller design [33] can be used [9]. First, recall notation used in description of the algorithm. In the constant spacing policy, the desired spacing between successive cars is defined by the formula x i,des = x i 1 L i, (15) where L i is a constant and includes the length l i 1 of the preceding vehicle. The spacing error of the i-th vehicle is ε i = x i x i 1 +L i (16) The subject of control is the acceleration of the i-th vehicle denoted as a i = ẍ i for i = 1,...,n. (17) Acceleration of the platoon leading vehicle is denoted asa 0. In the sliding surface method of controller design the following sliding surface is defined for our application: ω n S i = ε i + ξ + 1 ε i + C 1 (v i v 0 ), (18) ξ 2 11 C 1 1 C 1 where v i and v 0 are the i-th and leading vehicle velocities, respectively. If we set Ṡ i = λs i and λ = ω n (ξ + ) ξ 2 1 (19) one can prove that the desired acceleration of the i-th car is given by the formula a i,des = ẍ i,des = (1 C 1 )ẍ i 1 +C 1 ẍ 0 ( (2ξ C 1 ξ + )) ξ 2 1 ω n ε i ( ξ + ) ξ 2 1 ω n C 1 (v i v 0 ) ωn ε 2 i (20) As we can see in equation (20), the subject of tuning are the constants C 1, ξ and ω n. The gain C 1 is the weight of the preceding and leading car s accelerations and speeds. The gain ξ is the damping ratio and can be set to one for critical damping. The gain ω n is the bandwidth of the controller. It has been shown in [9] that if the cars in the platoon use control law (20), then they will be able to track the preceding car with a constant spacing. This means that the spacing error will converge to zero in the absence of acceleration of the leading vehicle. C. Predictive Cooperative Cruise Control Although the performance of the CACC algorithm is significantly better than the ACC performance, there is a room for further improvements. One of the possible improvements is to transmit desired acceleration values instead of the instantaneous ones. This modification does not change the CACC formula used, i.e. (20), but it has a significant impact on the platoon performance. To distinguish between these two algorithms, CACC and Predictive Cooperative Adaptive Cruise Control (pcacc) names will be used, respectively. In the remaining sections (including that describing the investigation results) the pcacc algorithm is used. IV. CAR AND PLATOON MODEL In this section we describe the platooning scenario and the model of cars used in our study. A. Actuation Lag Model The acceleration computed as the output of the ACC/CACC controller is not applied immediately due to the actuation lag. The lag is modelled as a low-pass filter applied to the output of the cruise controller: P(s) = 1 τs+1. (21) where τ is the time constant of the filter. A list of factors contribute to the actuation lag: the pure time delay in the engine response (60 ms at 2000 rotations per minute), the bandwidth of the lower level multiple-sliding surface controller that tracks acceleration, the bandwidth of low pass filters used for other sensors e.g. engine pressure sensor, wheel speed sensor, etc. the bandwidth of the throttle actuator, the lag due to discrete sampling at 50 Hz (20 ms), the 200 ms lag due to the radar filter, during braking, the brake actuator lag (up to 70 ms with ABS). Taking all these factors into account, we conclude that time constant τ could be as much as 500 ms. Therefore, the shortest time gap (headway time) which can be applied in ACC systems is 1 second. Thus, for vehicle speed of 30 m/s, having the time gap equal to 1 second, the distance between subsequent cars could be 30 meters. CACC reduces the distance considerably.

6 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 6 B. Model of Acceleration Constraints The acceleration capabilities have quite a significant impact on the behavior of cars in the platoon. In [34], two acceleration modelling approaches can be found: vehicle dynamics models and vehicle kinematics models. The vehicle dynamics models are very complex and require a significant number of input parameters. Therefore, we decided to choose a simple model from the vehicle kinematics family, namely a Linear Decay Model. In this model, the maximum achievable acceleration a at a given time and speed u is: a(t) = a m a m u e u(t) (22) where a m is the maximum possible acceleration rate and u e is the equilibrium speed, also known as the crawl speed. In our simulations we have chosen the values of a m and u e to be 4 m/s 2 and 210 km/h, respectively. We assume that, at any time and speed, the maximum deceleration is 0.3g (2.943 m/s 2 ), which corresponds to the limit of brake capacity of cruise control systems imposed by ISO 15622:2010 standard due to passenger comfort. C. Long Range Radar Model We assume that the distance and relative speed to the preceding car are measured using a Long Range Radar (LRR) that operates in 77 GHz band, such as Bosch LRR3 [35] or Continental ARS [36]. A Frequency Modulated Continuous Wave (FMCW) with the cycle time of 60 ms is used for the measurements. We assume that the distance and relative speed estimates are readily available at the end of the cycle. The accuracy of the measurements is assumed to be perfect, which is close to the reality based on the accuracies reported in [35], [36]. D. Platoon Assumptions The simulations are performed for the highway scenario with one lane occupied by a platoon of 10 cars. There are no other cars in the considered lane. There is a varying number of non-platooned cars in the remaining three lanes of the highway, set to 0, 5, or 10 cars/km/lane. Non-platooned cars move at the constant speed of 130 km/h. The platoon is preceded by a jamming vehicle that moves according to the predetermined speed pattern shown in Fig. 5. The jammer disrupts the fluent movement of the platoon and forces the platoon leader to adapt its speed. The jammer s speed changes in 30 s cycles, in which it decelerates with the factor of 0.3g m/s 2 until the speed of 30 km/h is reached. Then, it accelerates with the factor of 1.5 m/s 2 until the maximum speed of 130 km/h is reached. Finally, the jammer travels at the maximum speed until the start of the next cycle. The speed of the platoon leader is controlled by ACC based on the input from the long-range radar that monitors the distance to the jammer and its relative speed. All other cars within the platoon (followers) use the pcacc algorithm. This algorithm uses information transmitted via radio channel in addition to the long-range radar measurements. It is assumed speed [km/h] time [s] Fig. 5. Speed of the jamming vehicle that all the cars within the platoon broadcast their speed and acceleration information. E. Simulation Assumptions All results are obtained using a simulator that has been designed and implemented by the authors. The simulation tool supports various radio communication models but for the purpose of this paper we consider only the two types of communication: ideal and p. To investigate the platoon performance, the following Key Performance Indicator (KPI) are considered: inter-car distance - represents the "density" of the platoon. Shorter inter-car distance typically translates to higher traffic efficiency and fuel savings, packet reception rate - rate of the succesful packet reception for the packets transmitted by the platoon leader or preceding car, packet latency - latency of the received packets introduced by the channel contention, transmission time, propagation delay, and reception time. The aforementioned KPIs are the basis for the comparison of different ACC and CACC algorithms presented in following sections. V. COMMUNICATION REQUIREMENTS FOR CACC Here we study the impact of actuation lag and message rate (in Hz) on minimum feasible inter-cars spacing with ACC/CACC/pCACC. The study assumes ideal (zero-loss zerolatency) communication for CACC/pCACC. Every car periodically broadcast its speed and acceleration (CACC) or desired acceleration (pcacc). We then introduce a controlable message latency to study its impact on CACC/pCACC. A. Impact of Actuation Lags As presented in Section III, different vehicle controlling algorithms can be used to operate vehicles within the platoon. The considered algorithms are, however, prepared to be implemented in the real vehicles, so their behavior will be affected by non-ideal characteristics of the components that may introduce lags to the decision-making and actionperforming processes. Nowadays, the actuation lag of road vehicles can be up to 500 ms, as discussed in the previous section. Cars are

7 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 7 typically driven by humans, which have long reaction time, so there is no strong reason to reduce the actuation lag through optimization. However, we expect that in the era of automated driving, this lag will be substantially reduced (i.e. to few tens of miliseconds). In Fig. 6, we show the impact of 500, 100, and 20 ms actuation lags on the platoon performance. The mean inter-car distance is averaged over all pairs of consecutive cars and over time. The assumed message broadcast rate is 10 Hz. Mean inter-car distance [m] ACC CACC pcacc Fig. 6. Impact of the actuation lag on the platoon performance τ = 20ms τ = 100ms τ = 500ms The impact of the actuation lag strongly depends on the type of the cruise control used. The largest impact can be observed for the ACC and CACC algorithms. When the lag increases from 100 to 500 ms, the stable mean inter-car distance with ACC increases threefold (from 5.5. to 16 m). The distance achieved with CACC is somewhat shorter, but it is still strongly dependent on the actuation lag. As expected, pcacc is much less sensitive: It enables very short inter-car spacing ( 1 m) irrespective of the actuation lag. By using the desired accelerations of the preceding vehicle and platoon leader as inputs ( instead of their actual/current accelerations), pcacc is able to predict what could be the actual accelerations of the preceding vehicle and platoon leader after the actuation lag. In that way, the actuation lag is removed from the control loop. B. Impact of Message Rate The impact of the message broadcast rate on CACC and pcacc is show in Fig. 7 with the actuation lag equal to 500 ms. For the rates of 1 and 2 Hz, the performance of the platoon is significantly decreased. This is due to the fact that these frequencies correspond to 1 s and 500 ms information update intervals, which at highway speeds translate to long traveling distance without information update. For rates above 8 Hz, the performance degradation is rather small. For the rates above 16 Hz, the platoon performance improvement is insignificant, thus increased usage of radio resources has no justification. We therefore conclude that, for the considered scenario, the optimal information update rates are from 8 to 16 Hz. Mean inter-car distance [m] CACC pcacc 1Hz 2Hz 5Hz 8Hz 10Hz 16Hz 20Hz CACC message frequency Fig. 7. Impact of the CAM message rate on the platoon performance C. Impact of Message Latency To study the impact of the message delay, we introduce artificial delays of 25, 50, 75 and 100 ms into our ideal communication model. The results shown in Fig. 8 assume actuation lag of 500 ms and message rate of 10 Hz. Mean inter-car distance [m] CACC pcacc 0ms 25ms 50ms 75ms 100ms Communication delay Fig. 8. Impact of the artificial packet latency on the platoon performance With CACC, which requires much longer inter-car distance than pcacc, the message delay has a moderate impact. For the longest considered delay of 100 ms, the spacing increases by 14% w.r.t. ideal (zero-latency) communication. In the case of pcacc, the relative increase is much larger: the delay of 25 ms increases the inter-car distance by 73 % w.r.t. ideal communication. This indicates that pcacc algorithm is vulnerable to channel access delay caused by channel congestion in case of p communication. VI. PERFORMANCE OF PCACC WITH IEEE P COMMUNICATION In this section, we investigate the performance of pcacc supported by IEEE p communication. We investigate both the single-radio and dual-radio configurations. The dualradio configuration enables multi-channel operation.

8 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 8 A. Background on IEEE p and Standards IEEE p is one of the approved amendments to the IEEE standard to add Wireless Access in Vehicular Environments (WAVE) required to support applications of Intelligent Transportation Systems (ITS). It is complemented with IEEE , which specifies multi-channel operation of p. Together they are used as Physical (PHY) and Medium Access Control (MAC) layers of the Dedicated Short- Range Communications (DSRC) protocol stack. DSRC operates on seven channels (one control and six service channels) shown in Fig. 9, each of which is 10 MHz wide. Frequency [GHz] Channel number Channel usage Service (security) Service Service Fig. 9. Frequency allocation of p channels Service Service Service (security) Channel CH GHz is the control channel responsible for controlling the transmission broadcast and link establishment. CH GHz and CH GHz are service channels dedicated to traffic safety applications. The remaining four service channels are available for bidirectional communication between different types of transceivers [37]. Every DSRC-equipped vehicle broadcasts Basic Safety Messages (BSMs) on CH172 every 100 ms. The format of BSMs is specified in SAE J2735 standard. A BSM consists of two parts: the first part - mandatory - contains current information about the transmitting vehicle: position, movement trajectory, velocity, acceleration, brakes status, etc. the second part - optional - that may contain any additional information that is not explicitly specified in the standard (e.g. the desired acceleration used as an input to the pcacc algorithm). CACC may rely on BSMs. The equivalent of the BSM in the ETSI C-ITS standard is callled Context Awarnes Message (CAM). The MAC protocol in IEEE p uses the Enhanced Distributed Channel Access (EDCA), which is an enhanced version of the Distributed Coordination Function (DCF) from [38]. EDCA uses Carrier sense multiple access with collision avoidance (CSMA/CA), where a node that has data to transmit needs to perform medium sensing first. B. Simulation Assumptions In this work, we investigate a scenario when a platoon of ten cars is travelling on the outer lane of a four-lane highway (two lanes in each direction), while a number of non-platooned cars are travelling on the remaining three lanes that are not occupied by the platoon. The number of non-platooned cars is a simulation parameter. All cars on the highway transmit Basic Safety Messages (BSMs) with the frequency of 10 Hz on channel CH172. We investigate two transceiver configurations for platooned cars: single-radio transceiver that is continuously tunned to CH172 to transmit and receive BSMs, dual-radio transceiver, where the first radio is continuously tunned to CH172 to transmit and receive BSMs and the second radio is tunned to another Service Channels (SCHs). We do not consider multi-channel operation with the single-radio transceiver (i.e. channel switching), as according to [12], higher collision rates and increased message delays might be experienced in the corresponding alternating mode. Moreover, according to the SAE specification [39], a vehicle should continuously listen to CH172 for BSMs, thus prohibiting the use of this mode. In order to increase the efficiency of pcacc, we assume that besides BSMs, every platooned vehicle may broadcast short CACC packets, which contain only speed and acceleration, on a dedicated SCH. This is possible only with dual-radio transceiver configuration since the singleradio transceiver is continuously tunned to CH172. The transmission frequency of CACC packets is assumed to be 10 Hz. Other simulation parameters and assumptions are summarized in Table I and are similar to those presented in Section IV, except for the actuation lag, which is set to 20 ms to account for the future optimizations in the era of self-driven vehicles. TABLE I SELECTED SIMULATION PARAMETERS Parameter Assumption Number of simulation runs 100 Simulation time 900 s Platoon size 10 cars Actuation lag 20 ms ACC headway time 0.2 s BSM message size 300 bytes CACC message size 16 bytes We consider the following densities of non-platooned cars: 0 cars/km/lane - denoted as (4;0), 5 cars/km/lane - denoted as (4;5), 10 cars/km/lane - denoted as (4;10). C. pcacc Target Distance Selection The first step in the evaluation aims to determine the minimum feasible target distance between platooned cars. The target distance is the input parameter of the pcacc controller. We assume that the minimum feasibe target distance is the one that provides crash-free platoon operation with 99%- probability. Namely, we performed 100 simulation runs for different target distances ranging from 0.2 m to 15 m and then selected the minimum distance for which no more that one out of 100 simulations ended with a crash to be the minimum feasible target distance. Our simulator does not implement any crash mitigation mechanisms (i.e. emergency braking) that would mitigate hazardous situations in realworld implementations. CACC alone is not ment to deal with

9 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 9 such situations. Therefore, we assume that 100% crash-free operation is not an imperative in our simulations, but 99% target is sufficient Ideal - single Ideal - dual (4;0) single (4;0) dual (4;5) single (4;10) single (4;5) dual (4;10) dual simulation time, and over all simulation runs that ended without crash. The mean inter-car distance is somewhat higher than the target distance of the pcacc controller. This may be the consequence of the actuation lag and the imposed acceleration constraints, as they limit the ability of vehicles to maintain the target distance. However, one can state that pcacc operates according to the provided input. Crash ratio Mean Platoon Inter-car Distance CACC target distance [m] Fig. 10. Fraction of simulations concluding with a car crash per 100 runs. Mean inter-car distance [m] From the results presented in Fig. 10 and Table II we can see that in all-but-two considered scenarios it is possible to find such a minimum feasible target distance in the 0.2 m to 15 m range. In the two scenarios, namely (4;5) and (4;10) with the single-radio transceiver configuration, we selected the target distance that results in the smallest number of crashes. In the rest of this section, we present the results obtained for the target distance indicated in Table II. TABLE II SELECTED PCACC TARGET DISTANCES AND THE CORRESPONDING NUMBER OF SIMULATIONS WITH A CRASH PER 100 RUNS. Case Target distance [m] No. of crashes Ideal, single Ideal, dual (4;0) p, single (4;0) p, dual (4;5) p, single (4;5) p, dual (4;10) p, single (4;10) p, dual One should note that the use of the dual-radio transceiver and special CACC messages transmitted on the dedicated SCH channel results in substantial improvement of pcacc performance. The minimum feasible target distances achieved in such configuration are always better than in case of a singleradio. This is the result of increased overall message rate with the dual-radio, which is 20 Hz (10 Hz BSM on CH172 plus 10 Hz CACC on SCH), as well as the improved message reception ratio: While BSM packets might collide frequently, as nonplatooned cars also use CH172 for BSM transmissions, the reliability of CACC messages is much higher because they are transmitted on a dedicated SCH channel. D. Results Here we present the results obtained for the target distances listed in Table II. Fig. 11 shows the mean inter-car distance in the platoon for all considered scenarios. The mean is calculated over all consecutive pairs of cars, over the whole Ideal - single Ideal - dual (4;0) single (4;0) dual (4;5) single (4;5) dual (4;10) single (4;10) dual Case Fig. 11. Observed mean inter-car distance in platoon in a single-lane scenario. To analyze the realiability of p message transmissions, we collected the message reception statistics at each car in the platoon. The reception rate of the messages transmited by the platoon leader are shown in Fig. 12. The reception rate of the messages transmited by the preceeding car are shown in in Fig. 13. One may notice from Fig. 12 that the reception rate of leader messages drops significantly towards the tail of the platoon when the single-radio is used. The last car in the platoon was unable to receive more than 2% of messages due to low reliability of p communication with the leader. If several consecutive packets are lost, the pcacc algorithm might be unable to adjust the speeds of the cars properly and, as a result, they may crash with each other. Different situations can be observed when dual-radio is used. In this case, the higher transmission reliability of CACC messages may compensate for the loss of BSM messages to allow crashfree platooning with a shorter target distance, as hardly any CACC packets are lost. Another very important requirement of CACC is up-to-date information about the acceleration and velocity of platooned cars. Therefore, message latency, shown in Fig. 14 for BSMs and CACC packets, respectively, is an important KPI. One may notice that increase in density of non-platooned cars has very little impact on the latency of transmitted messages. There is also hardly any performance difference when comparing the single-radio and dual-radio transceivers. This is due to fact that BSM and CACC messages are broadcasted. Hence, no retransmissions are performed, and there is no increase in the EDCA backoff Contention Window (CW) size, which remains at its minimum value all the time. Consequently, low transmission latency is an advantage of the p communications protocol.

10 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 10 Reception rate Reception rate of leader packets Car (4;0) single (4;0) dual (4;5) single (4;5) dual (4;10) single (4;10) dual Fig. 12. Reception rate of leader packets vs. the car position in the platoon. Reception rate Reception rate of preceding car packets Car (4;0) single (4;0) dual (4;5) single (4;5) dual (4;10) single (4;10) dual Fig. 13. Reception rate of preceding car packets vs. the car position in the platoon. One should also note that the latency of CACC messages is less than half of the BSMs message latency. This is a consequence of CACC messages being much shorter than BSMs and the channel used for their transmission being less congested than CH172. Average latency [ms] (4;0) single (4;0) dual (4;5) single (4;5) dual (4;10) single (4;10) dual Case Fig. 14. Average latency of received p packets. BSM CACC We conclude from the presented results that pcacc can be effectively used for high-density platooning as long as reliable, low-latency wireless communications is provided. With p this can be achieved with the use of dualradio transceivers. The performance of pcacc with singleradio transceivers strongly depends on the traffic intensity on the highway, with even moderate traffic (5 cars/km/lane) resulting in significant channel congestion and message loss. An alternative to p could be the use of LTE V2V communication recently standardized by 3GPP. VII. CONCLUSIONS The article considered the impact of the communications requirements on the performance of a platoon of cars that are controlled with different vehicle controlling algorithms. The paper investigated both the ideal and the IEEE p based communication. It was shown, with the use of ideal communication that the impact of the actuation lag, communication delay and message rate is crucial for the platoon performance. Moreover, with the use of IEEE p based communication we evaluated the performance of the platoon in realistic highway scenarios. Obtained results reveal that performance of the platoon is strictly correlated with the applied vehicle control algorithm and the radio communication aspects. The best results are observed for the predictive CACC which outperforms the remaining algorithms significantly, regardless the actuation lag or communication delay. The platoon performance is also affected by the message frequency rate and the optimal value equal to 10 Hz. Lower and higher message frequency rates only deteriorates or have negligible impact on its performance. Interesting results can be also observed in the case of IEEE p communication. Results show that the performance of the platoon formed with the vehicles equipped with the dualradio radio transceivers is close to ideal regardless the highway traffic intensity. Moreover, one should note, that p based platoons experience very small packet latency what can be a crucial argument for low latency safety applications. ACKNOWLEGMENT This work has been performed by Poznan University of Technology in cooperation with Nokia Bell Labs, Wroclaw, Poland, within the contract No REFERENCES [1] J. Lioris, R. Pedarsani, F. Y. Tascikaraoglu, and P. Varaiya, Doubling Throughput in Urban Roads by Platooning, in Proc. IFAC Symposium on Control in Transportation Systems, Istanbul Turkey, [2] Scania. Annual Report 2001, [Online]. Available: scania.com/group/en/wp-content/uploads/sites/2/2015/09/sca01en_ tcm _tcm pdf. [3] SARTRE Project, D.4.3 Report on Fuel Consumption, [Online]. Available: SARTRE_4_003_PU.pdf, [4] J. Ploeg, B. Scheepers, E. van Nunen, N. van de Wouw, and H. Nijmeijer, Design and Experimental Evaluation of Cooperative Adaptive Cruise Control, in Proc. IEEE Int. Conf. Intelligent Transportation Systems (ITSC), Washington, USA, [5] R. Rajamani, H.-S. Tan, B. K. Law, and W.-B. Zhang, Demonstration of Integrated Longitudinal and Lateral Control for the Operation of Automated Vehicles in Platoons, IEEE Trans. Control Systems Tech., vol. 8, pp , [6] S. Santini, A. Salvi, A. S. Valente, A. Pescape, M. Segata, and R. L. Cigno, A Consensus-based Approach for Platooning with InterVehicular Communications, in Proc. IEEE Conf. Computer Comm. (INFO- COM), Hong Kong, 2015.

11 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 11 [7] D. Jia, K. Lu, and J. Wang, A Disturbance-Adaptive Design for VANET-Enabled Vehicle Platoon, IEEE Trans. Veh. Tech., vol. 63, pp , [8] D. Jia, K. Lu, J. Wang, X. Zhang, and X. Shen, A Survey on Platoonbased Vehicular Cyber-Physical Systems, IEEE Comm. Surveys and Tutorials, vol. 18, pp , [9] R. Rajamani, Vehicle Dynamics and Control. Springer, [10] J. B. Kenney, Dedicated Short-Range Communications (DSRC) Standards in the United States, Proc. IEEE, vol. 99, pp , [11] A. Festag, Cooperative Intelligent Transport Systems (C-ITS) Standards in Europe, IEEE Comm. Mag., vol. 12, pp , [12] IEEE Standard for Wireless Access in Vehicular Environments (WAVE) - Multi-Channel Operation, January 2016, in IEEE Std (Revision of IEEE Std ). [13] S. Tsugawa, S. Kato, and K. Aoki, An Automated Truck Platoon for Energy Saving, in 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems, Sept 2011, pp [14] European Truck Platooning Challenge, [Online]. Available: [15] Volvo Trucks Successfully Demonstrates On-Highway Truck Platooning in California, [Online]. Available: about-volvo/news-and-events/, [16] C. Lei, E. M. van Eenennaam, W. K. Wolterink, G. Karagiannis, G. Heijenk, and J. Ploeg, Impact of Packet Loss on CACC String Stability Performance, in th International Conference on ITS Telecommunications, Aug 2011, pp [17] M. Segata, B. Bloessl, S. Joerer, C. Sommer, M. Gerla, R. Lo Cigno, and F. Dressler, Towards Communication Strategies for Platooning: Simulative and Experimental Evaluation, IEEE Transactions on Vehicular Technology, vol. 64, no. 12, pp , December [18] P. Fernandes and U. Nunes, Platooning with IVC-Enabled Autonomous Vehicles: Strategies to Mitigate Communication Delays, Improve Safety and Traffic Flow, IEEE Transactions on Intelligent Transportation Systems, vol. 13, no. 1, pp , March [19] A. Bohm, M. Jonsson, and E. Uhlemann, Performance Comparison of a Platooning Application Using the IEEE p MAC on the Control Channel and a Centralized MAC on a Service Channel, in 2013 IEEE 9th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), Oct 2013, pp [20] M. Boban and A. Festag, Service-actuated Multi-channel Operation for Vehicular Communications, Computer Communications, vol. 93, pp , 2016, multi-radio, Multi-technology, Multi-system Vehicular Communications. [Online]. Available: science/article/pii/s [21] N. Lyamin, A. Vinel, and M. Jonsson, Does ETSI Beaconing Frequency Control Provide Cooperative Awareness? in 2015 IEEE International Conference on Communication Workshop (ICCW), June 2015, pp [22] P. Ioannou and C. C. Chien, Autonomous Intelligent Cruise Control, IEEE Trans. Veh. Technol., vol. 42, no. 4, pp , [23] A. D. Ames, J. W. Grizzle, and P. Tabuada, Control Barrier Function based Quadratic Programs with Application to Adaptive Cruise Control, in Proc. IEEE Conference on Decision and Control, [24] A. Mehra, W.-L. Ma, F. Berg, P. Tabuada, J. Grizzle, and A. D. Ames, Adaptive Cruise Control: Experimental Validation of Advanced Controllers on Scale-Model Cars, in Proc. American Control Conference (ACC), [25] C. Wang and H. Nijmeijer, String Stable Heterogeneous Vehicle Platoon Using Cooperative Adaptive Cruise Control, in Proc. IEEE 18th Intern. Conf. on Intelligent Transportation Systems, 2015, pp [26] J. Ploeg, E. Semsar-Kazerooni, G. Lijster, N. van de Wouw, and H.Nijmeijer, Graceful Degradation of CACC Performance Subject to Unreliable Wireless Communication, in Proc. IEEE 16th Intern. Annual Conf. on Intelligent Transportation Systems, 2013, pp [27] P. Wang, Z. Sun, J. Tan, Z. Huang, Q. Zhu, and W. Zhao, Development and Evaluation of Cooperative Adaptive Cruise Controllers, in Proc. IEEE Intern. Conf. on Mechatronics and Automation, [28] F. Morbidi, P. Colaneri, and T. Stanger, Decentralized Optimal Control of a Car Platoon with Guaranteed String Stability, in Proc. European Control Conference (ECC), [29] J. Ploeg, Analysis and Design of Controllers for Cooperative and Automated Driving, Ph.D. dissertation, Technische Universiteit Eindhoven, [30] D. Yanakiev, J. Eyre, and I. Kanellakopoulos, Analysis, Design, and Evaluation of AVCS for Heavy-duty Vehicles with Actuator Delays, California PATH research report, UCB-ITS-PRR-98-18, [31] F. Bu, H.-S. Tan, and J. Huang, Design and Field Testing of a Cooperative Adaptive Cruise Control System, in Proc. American Control Conference (ACC), 2010, pp [32] J. Mårtensson, A. Alam, S. Behere, M. A. A. Khan, J. Kjellberg, K.- Y. Liang, H. Pettersson, and D. Sundman, The Development of a Cooperative Heavy-duty Vehicle for the GCDC 2011: Team Scoop, IEEE Trans. Intell. Transp. Syst., vol. 13, no. 3, pp , [33] J. J. E. Slotine and W. Li, Applied Nonlinear Control. Prentice Hall, [34] H. Rakha, M. Snare, and F. Dion, Vehicle Dynamics Model for Estimating Maximum Light-duty Vehicle Acceleration Levels, Transportation Research Record: Journal of the Transportation Research Board, vol. 1883, pp , [35] Bosch. LRR3: 3rd generation Long-Range Radar Sensor. [Online]. Available: db_application/pdf_2/en/lrr3_datenblatt_de_2009.pdf. [36] Continental. ARS Premium Long Range Radar Sensor 77 GHz. [Online]. Available: download/industrial_sensors_de_en/themes/download/ars_premium_ datenblatt_en.pdf. [37] Y. Li, An Overview of the DSRC/WAVE Technology, in Security and Robustness in Heterogeneous Networks, QShine 2010, Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering, X. Zhang and D. Qiao, Eds. Springer, 2012, vol. 74. [38] IEEE Standard for Information technology Telecommunications and information exchange between systems local and metropolitan area networks Specific requirements Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications, March [39] On-Board System Requirements for V2V Safety Communications, March Michał Sybis received the M.Sc. in electronics and telecommunications and Ph.D (with honors) degree in telecommunications from the Poznan University of Technology, Poznan, Poland in 2007 and 2012, respectively. He is currently an Assistant Professor with the Chair of Wireless Communications at the Faculty of Electronics and Telecommunications at Poznan University of Technology. His research interests include coding techniques, iterative decoding, the fifth generation mobile wireless systems and vehicular communications. Vladimir Vukadinovic is a senior research engineer at Nokia. He received his Dipl.-Ing. and M.Sc. degrees in Telecommunications from the University of Belgrade, Serbia in 2000 and 2003, respectively, and his Ph.D. degree in Telecommunications from the Royal Institute of Technology (KTH), Stockholm, in Before joining Nokia, he was a senior research scientist at the Institute of Networked and Embedded Systems, University of Klagenfurt and an associate research scientist at the Disney Research, Zurich. His research interests include M2M communication, V2X for ITS, radio resource management and video streaming protocols.

12 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS 12 Marcin Rodziewicz is a Teaching Assistant at the Poznan University of Technology. He received his M.Sc. degree in telecommunications from the Poznan University of Technology in He is currently working towards the Ph.D. degree in telecommunications at the same university. His research interests include D2D and V2V communications, radio resource management and power control mechanisms. Paweł Sroka received the M.Sc. and Ph.D. (with honors) degrees in telecommunications from Poznan University of Technology (PUT), Poland, in 2004 and 2012, respectively. Currently he is employed as an Assistant Professor at the Chair of Wireless Communications, PUT. For several years he has participated in various international and national research projects. His main research interests include radio resource management for wireless networks, cross-layer optimization and MIMO systems. Krzysztof Wesołowski has been employed at Poznan University of Technology (PUT), Poznan, Poland, since He received the Ph.D. and Doctor Habilitus degrees in communications from PUT in 1982 and 1989, respectively. Since 1999 he has held the position of Full Professor in telecommunications. Currently he is Dean of the Faculty of Electronics and Telecommunications at PUT. In his scientific activity he specializes in digital wireline and wireless communication systems, information and coding theory and DSP applications in digital communications. He is the author or co-author of more than 150 scientific publications, including the books: "Mobile Communication Systems", John Wiley & Sons, Chichester 2003, and "Introduction to Digital Communication Systems", John Wiley & Sons, Chichester, He published his results, among others, in IEEE Transactions on Communications, IEEE Journal on Selected Areas in Communications, IEEE Transactions on Vehicular Technology, IEE Proceedings, European Transactions on Telecommunications, Electronics Letters and EURASIP Journal on Wireless Communications and Networking. Professor Wesołowski was a Postdoctoral Fulbright Scholar at Northeastern University, Boston, and a Postdoctoral Alexander von Humboldt Scholar at the University of Kaiserslautern, Germany. He also worked at the University of Kaiserslautern as a Visiting Professor. His team participated in several domestic and international research projects funded by industry and the European Union within the Sixth and Seventh Framework Programs. Adrian Langowski is an Assistant Professor at Poznan University of Technology. He received his M.Sc degree in telecommunications from the Poznan University of Technology in 2001, and his Ph.D. degree in radio communications from the Poznan University of Technology in From the beginning of his career he has been working at the Poznan University of Technology. His current research interests include M2M communications, especially V2X and communication between UAVs. Karolina Lenarska received her M.Sc. degree in electronics and telecommunications in 2012 and B.Sc. degree in computer science in 2013, both from the Poznan University of Technology (PUT). She has been employed at PUT since October 2012 as a Senior Researcher, and since October 2015 as a Teaching Assistant. Currently she is working towards the Ph.D. degree and her research area covers the fifth generation mobile wireless systems, V2V communication and multiple antenna techniques, such as beamforming and interference alignment.