Enhancing ETSI DCC for Multi-Service Vehicular Safety Communication Mohammad Irfan Khan † , Jérôme Härri † , Stefania Sesia ‡ † EURECOM, Sophia-Antipolis, France. E-mails: {khanm,haerri}@eurecom.fr ‡ Renault SAS, Sophia-Antipolis, France. E-mail: {stefania.sesia}@renault.com Abstract—ETSI Decentralized Congestion Control (DCC) for vehicular communication is an essential mechanism for limiting wireless channel congestion and resource allocation of ad-hoc V2X communications. Standardized channel congestion control protocols have been designed considering mostly a single message for cooperative awareness such as CAM/BSM, while future automated vehicles will exchange additional messages, including sensor data, control information, HD-maps etc. In this paper we evaluate and improve the performance of the state-machine based DCC standardized in Europe by ETSI. We highlight the channel capacity under-utilization and the communication quality degradation, and propose three design improvements to enhance its performance. Simulation based evaluation, considering multiple standardized safety messages on a single channel, proves the performance improvement due to our proposed modiﬁcations, almost doubling the reception throughput for dense V2X communication scenarios. I. I NTRODUCTION Vehicle-to-Everything (V2X) communication will be soon ubiquitous on our roads to improve road safety and increase trafﬁc efﬁciency by increasing a vehicle’s awareness by com- municating information beyond the driver’s visual range and the vehicle’s on-board sensors. Over the years, V2X net- working protocols and communication technologies have been consolidated and is currently available for initial a.k.a Day 1 deployment. Two leading technologies have been developed for V2X communication, i.e. IEEE 802.11 based ITS-G5 in Europe/DSRC in the USA, and 3GPP Long-Term-Evolution (LTE) V2X. In IEEE 802.11 based vehicular networks and Mode 4 of LTE-V2X, there is no centralized channel resource allocator and the nodes need to prevent channel saturation by limiting the spatial and/or temporal channel usage through cooperative strategies. Several wireless channel congestion control protocols have been proposed by the academia based on Transmit Rate Control (TRC) [1]–[3] and Transmit Power Control (TPC) [4], [5], while some protocols have been standardized for Day 1 deployment such as [6], [7]. In European standards developed by ETSI, TRC has been speciﬁed as the principle mechanism for congestion control at the Access layer [6], a.k.a Decentralized Congestion Control (DCC). Whether academia or standardization, the focus has been mainly to optimize channel usage while limiting Channel Load (CL) considering a single periodic broadcast message such as Cooperative Awareness Message (CAM) or Basic Safety Message (BSM). In future or Day 2 scenarios, revolutionary V2X applica- tions, such as highly automated driving (HAD) and safety of Vulnerable Road Users (VRU) will be based on a multitude of V2X services, such as Collective Perception (CP) [8] and Maneuver Coordination (MC) [9]. Several types of messages, such as CPM and MCM will be generated by such applica- tions, which will strain the communication resources. In this regard, ITS-G5 has been criticized for having insufﬁ- cient capacity. However, previous studies [10]–[12] and ETSI Technical Report TR 103 562 [8] have demonstrated that the inefﬁcient channel usage with regards to DCC is a problematic aspect of ITS-G5. Accordingly, complementary to developing multi-channel operations or new radio technologies to acquire more channel capacity, it is also critical to use the channel in the most efﬁcient way. In a previous study [10], we demonstrated the inefﬁciency of the initial version of DCC (v 1.1.1, 2011 [13]) with multiple types of safety V2X messages. In this paper we demonstrate the shortcomings of a recently revised version of DCC (v1.2.1, 2018 [6]) and propose 3 modiﬁcations, i.e i) Shifting the approach from transmit (Tx) rate control to channel resource control, ii) Setting less severe rate control parameters to maximize channel usage, iii) Adopting continuous and smooth adaptation, instead of abrupt state machine based rate transi- tion. These modiﬁcations vastly contribute to ITS-G5 using higher channel capacity, while limiting the channel congestion. The rest of the paper is organized as follows: Section II presents a brief overview of standardized ETSI DCC, while Section III analyzes the issues and challenges of the Reactive variant of DCC. Section IV presents our proposed improve- ments, followed by Section V which provides simulation based performance evaluation results. Finally Section VI concludes the paper. II. ETSI ACCESS DCC - BRIEF OVERVIEW &ANALYSIS ETSI Access Layer DCC has been speciﬁed in the standard ETSI TS 102 687 [6], which describes the TRC mechanism. The input to the TRC algorithm is the Channel Busy Ratio (CBR), calculated as the proportion of time the wireless channel is sensed busy. The output is the Transmit Rate Limit (TRL) or the minimum Inter Transmit Time (ITT) a.k.a Toff, an obligatory interval of non-transmission after each transmission, enforced via trafﬁc shaping at the MAC layer. The Toff is enforced either using a Reactive mechanism via table lookup or an Adaptive mechanism. Reactive DCC: Reactive DCC obtains the Toff from a lookup table for each range of CBR, as shown in Table I. Each range of CBR corresponds to a DCC state and the variation of the maximum Tx rate w.r.t CBR works as a state machine, as shown in Figure 1. When DCC is in the relaxed state, the highest Tx rate is 20 Hz for a CL < 30%, while the lowest rate is 1 Hz for CL > 65%, for the restrictive state. The states in between are Active states, further divided into states 1, 2 and 3. Transition is only possible between adjacent DCC