I Introduction
UltraReliable LowLatency Communications (URLLC) is a novel traffic type that will be supported by the nextgeneration cellular networks (5G) in addition to enhanced Mobile Broadband (eMBB) and massive MachineType Communications (mMTC) [1]. Many applications, such as autonomous vehicles interaction or telesurgery, generate uplink URLLCtraffic [2]. In particular, each autonomous vehicle transmits information collected from its sensors, e.g., information about its position, speed, acceleration, or obstacles detected on the road. Similarly, during the telesurgery, a surgeon transmits commands to remote robotic manipulators that react to these commands with corresponding actions. To ensure the efficient work of such applications, they impose very strict latency (several milliseconds) and reliability (higher than ) requirements [3]. It should be noted that problem of quality of service provisioning for these applications is not only considered for cellular networks, but also attracts much attention from the wired and WiFi communities [4]. In this paper, we focus on solution for 5G cellular networks.
The 3rd Generation Partnership Project (3GPP) specifications, e.g., [5], for 5G networks describes two methods for an uplink channel access: grantbased and grantfree (configured grant). According to the grandbased method, to access an uplink channel, a user equipment (UE) transmits a scheduling request and waits for a scheduling grant from a base station (called gNB). When the grant is received, the UE starts transmitting data. As a result, the grantbased channel access delays the actual transmission, which makes this method inapplicable to most of URLLC use cases. Hence, the grantfree channel access is usually considered for URLLC data transmission.
With the grantfree channel access, a gNB selects the channel resources and the transmission parameters, e.g., Modulation and Coding Scheme (MCS), for each UE in a longterm time scale. To improve reliability, UEs can perform multiple transmission attempts. Before the transmission attempt, a UE can either wait for the feedback from a gNB related to the previous attempts or perform the transmission without waiting for feedback. In this paper, we consider the scheme without feedback called Krepetition, which implies that a UE makes transmission attempts for each data packet using the parameters configured by the gNB. Since the Krepetition does not require receiving feedback from the gNB after each transmission attempt, it allows significantly reducing the data transmission latency. To improve the reliability, the gNB uses Hybrid Automatic Repeat reQuest (HARQ) scheme, e.g., Chase Combining (CC) [6]. Many works compare Krepetition with other schemes that use the feedback from a gNB using analytical models [7, 8] and simulations [9, 10]
. Their results show that Krepetition is more effective in terms of transmission latency and reliability except for the cases of high load with massive overlapping of transmissions, i.e., when transmissions corresponding to different UEs with high probability use the same channel resources.
Since the channel quality can significantly change in time and frequency domains, and the gNB selects the transmission parameters for relatively long periods of time, a UE should use robust transmission parameters to satisfy strict URLLC quality of service requirements. In this case, the minimization of channel resource consumption is challenging because the more robust parameters are used, the more channels resources are consumed, which leads to lower network capacity, i.e., the number of data flows that can be served simultaneously. Studies [7, 8, 9, 10] consider usage of fixed MCS and the number of transmission attempts, so they do not consider the dynamic parameter selection problem. The papers [11, 12] studies the selection of
to maximize the number of users in the network. They consider that in the case when the same resources are assigned to multiple UEs, simultaneous transmissions of different UEs may lead to unsuccessful transmissions. However, in both works, the probability of unsuccessful transmission is estimated without taking into account that the probability of unsuccessful transmission depends on the MCS.
The channel resource allocation problem for Krepetition is considered in [13]. The authors propose to divide available channel resources into multiple groups and randomly select a group for each transmission attempt. They propose to select MCS corresponding to the number of these groups (higher number of groups corresponds to higher code rates). This study shows that the Krepetitions scheme can provide the required packet loss probability with the considered in the paper transmission parameters values. However, the authors do not provide any adaptive method for selecting these parameters, e.g., depending on the channel quality. The paper [14] provides the algorithm for selecting and the channel resources for transmission. However, this work considers fixed probabilities of successful decoding without taking into account that they depend on the used MCS. The paper [15] provides the number of transmission attempts (i.e., ) selection algorithm based on an estimation of fading correlation function. The authors suggest using two attempts in the case of low correlation and four in the case of high correlation. Numerical results show that this method can increase reliability and resource efficiency in a multiuser scenario compared to using a fixed number of attempts. However, this work does not consider the problem of selecting MCS and proposes to use a fixed robust MCS.
The provided above literature analysis show that the existing papers do not provide any adaptive method for selecting MCS depending on channel conditions. In this paper, we propose the adaptive transmission parameters (i.e., MCS and number of transmission attempts) selection algorithm based on estimating of the packet loss ratio for each parameter configuration using the SignaltoNoise Ratio (SNR) statistics at the gNB.
Ii Proposed algorithm
In this paper, we propose the transmission parameters selection algorithm that allows a gNB to select MCS and the number of transmission attempts for a UE transmitting URLLC packets in uplink.
Let us consider a UE served by a gNB. The gNB configures the parameters for the UE, i.e., the resources that are available for its uplink transmissions and the transmission parameters, i.e., the MCS and the number of transmission attempts . To change the parameters configuration at the UE, the gNB transmits to the UE the configuration messages that include the new MCS and the number of transmission attempts. In its turn, the UE updates the configuration upon the reception of the new parameters. The timefrequency resources assigned to the UE are divided into multiple groups called Resource Block Groups (RBGs). When the UE transmits a packet, it selects an appropriate number of RBGs corresponding to the selected MCS and the packet size. To avoid correlated errors in the consecutive RBGs, we assume that the UE selects RBGs uniformly spaced in the frequency domain. These RBGs are selected independently for each attempt. The time slots for repeated attempts are also selected uniformly spaced within the packet delay budget to avoid correlated errors in time.
Initially, after the connection establishment, the gNB configures the most robust uplink transmission parameters (MCS 0 and , where is the maximum number of transmission attempts) because there is no actual SNR statistics for this UE. Based on Sounding Reference Signals (SRS) periodically transmitted by the UE, the gNB estimates the average SNR for each RBG. Then, for each combination of the transmission parameters MCS and (), the gNB estimates the BLock Error Rate (BLER), i.e., the error probability of a single transmission attempt, as follows:

The gNB calculates the number of RBGs that is required for transmission using the considered MCS. Here we assume that the gNB has information about the packet size (e.g, this information can be provided via crosslayer interaction between the gNB and the URLLC application [16]). Then, the gNB randomly selects RBGs that are uniformly spaced in the available bandwidth.

We assume that the total power remains the same regardless of the number of used RBGs since the UE uses the whole power to transmit data. Hence, the gNB recalculates SNR according to
where is a measured SNR in RBG , is a total number of RBGs in which the UE transmits SRS.

The gNB uses the Exponential Effective Signaltonoise ratio Mapping (EESM) [17] error model for BLER estimation.
Let us describe the third step in more detail. According to the EESM model, the gNB maps vector of SNRs for the selected RBGs to a single effective SNR as follows:
where SNR is the SNR value in the th RBG, is the set of allocated RBGs, and is a scaling parameter. We use values obtained in [17].
We assume that the gNB uses CC to decode several HARQ transmissions. The effective SNR after transmission attempts is calculated as follows:
where is the set of used RBGs for th transmission attempt (note, that ), SNR is the SNR experienced in .
The obtained effective SNR value is mapped to the BLER value using the SNRBLER curves for the corresponding MCS. In this study, we use SNRBLER curves (Fig. 1) obtained for 25 iterations of a decoder for packet size 32 byte and physical layer parameters described in Section IIIA.
The packet loss ratio after all transmission attempts equals , where BLER is the obtained BLER for the th transmission attempt.
As a result, after each SRS reception, the gNB obtains PLR estimations for all the configurations {MCS, }. For each configuration, the PLR estimation is averaged with an exponentially weighted moving average as follows:
where is the window size, is the averaged PLR estimation after the th SRS reception.
To avoid frequent reconfiguration, i.e., transmission parameters changes at the UE, the gNB uses two thresholds, PLR and PLR, where PLR PLR. Specifically, the gNB selects the transmission parameters as follows:

marks all configurations {MCS, } as valid, if , and as invalid, if (see Fig. 2);

calculates the channel resource consumption for each valid configuration and selects the one that provides the minimum channel resource consumption.
If the selected configuration differs from the one used by the UE, the gNB sends the new configuration to the UE.
Iii Performance evaluation
Iiia Scenario
We evaluate the performance of the proposed algorithm with the NS3 simulator [18]. We consider a single gNB and a single UE that generates Constant Bit Rate (CBR) traffic in the uplink. Specifically, bytes packets are transmitted with periodicity 10 ms. Each packet should be delivered within the ms time interval with a probability higher than . SRSs are transmitted every ms.
Let us describe physical layer parameters. Following [19], we consider minislots (subslots) that consist of two Orthogonal FrequencyDivision Multiplexing (OFDM) symbols corresponding to control and data channels. The duration of the OFDM symbol equals that corresponds to the kHz interval between subcarriers. So, there are 14 slots within the packet delay budget. The bandwidth equals MHz and consists of RBGs [5]. The UE transmission power equals dBm, and it is equally distributed between the selected RBGs. We summarize all simulation parameters in Table I.
Parameter  Value 

Bandwidth  100 MHz, 16 RBGs 
Slot length  
Packet size  32 bytes 
Packet period  10 ms 
SRS period  5 ms 
UE power  23 dBm 
UE height  1.5 m 
gNB height  30 m 
Propagation model  OkumuraHata model [20] 
Fading model  Extended Pedestrian A (EPA) [21] 
Simulation time  100000 s 
4 
IiiB Analysis of the results
As a characteristic of the average channel quality, which decreases with the distance between the UE and the gNB, we use the value called the wideband SNR. This value is calculated as follows: , where is a UE transmission power, is the available bandwidth, is the pathloss, is a noise power spectral density at the gNB. To model the channel quality changes in time and frequency domains, we use the Extended Pedestrian A (EPA) fading model [21].
First, let us estimate the reduction of resource consumption that can be achieved with the adaptive transmission parameters selection. For each value and each parameter configuration {MCS, }, we carry out the full experiment and consider the PLR and the average number of used RBGs obtained in each experiment. Then, for each , we select the optimal configuration that allows satisfying the URLLC requirements while consuming the minimal number of RBGs. We compare the resource consumption provided by the proposed adaptive algorithm with the resource consumption provided by the optimal configuration and by the fixed selection of the most robust MCS, i.e., MCS 0 (see Fig. 3). The number of transmission attempts for MCS 0 equals the minimal number that allows satisfying the URLLC requirements. We can see that the proposed algorithm provides more than three times a reduction for the resource consumption in comparison with the MCS 0 selection. Moreover, the results for the proposed algorithm are close to optimal for dB. Since even with the most robust configuration the URLLC requirements cannot be satisfied at dB, we assume in the further experiments that the UE is located uniformly in the circle with center at gNB and radius corresponding to dB.
Now let us study how the parameters , PLR and PLR influence the efficiency of the proposed algorithm. For that, we vary the window size from 500 ms to 20000 ms and the thresholds PLR and PLR from to , respectively. For each window size value, we find combination of PLR and PLR that provides PLR less than for all SNR values and the minimum RBG usage averaged over SNR distribution. Fig. 4 shows how the average number of used RBGs depends on the window size for the proposed algorithm. To obtain the optimal and MCS 0 curves, we average the number of used RBGs, presented in Fig. 3, over the SNR distribution. According to the results, the number of used RBGs for the proposed algorithm converges to optimal when the window size increases. Moreover, the proposed algorithm allows reducing the resource consumption more than twice in comparison with the fixed selection of the MCS 0.
Fig. 5 shows PLR and PLR thresholds selected by the proposed algorithm. We can see that PLR does not depend on the window size and equals , which corresponds to URLLC reliability requirement. The threshold PLR increases with the window size and tends to PLR for large window size because the large window allows more conservatively estimating the packet loss ratio provided by different configurations, and thus it does not require to select the lower values for PLR.
According to Fig. 6, the usage of two thresholds allows the gNB to rarely reconfigure the transmission parameters. In particular, for s the parameters should be reconfigured once per seconds on average. The reconfiguration frequency further decreases when the window size increases.
Based on the results presented in Fig. 4 and Fig. 5, we propose the algorithm to select the parameters of the algorithm. Specifically, we propose to select the window size according to the environmental conditions, e.g., the UE mobility parameters, and then select the PLR according to Fig. 5. The threshold PLR should be set in accordance with the URLLC reliability requirement.
Iv Conclusion
In this work, we study the problem of uplink transmission parameter selection with the grantfree channel access. We propose the adaptive algorithm for the transmission parameters selection, i.e., MCS and the number of transmission attempts. This algorithm allows satisfying the URLLC requirements while reducing the channel resource consumption. The algorithm uses the signaltonoise ratio statistics to estimate the packet loss ratio for all possible transmission parameter values. Then the algorithm selects the parameter values that requires the minimum amount of channel resource while satisfying the URLLC requirements. Numerical results obtained with the NS3 simulator show that the algorithm reduces the channel resource consumption more than twice in comparison with the fixed robust MCS and optimal K parameters selection. Moreover, with this algorithm, the gNB can select the transmission parameters in a longterm perspective, e.g., it can change the parameters for the time intervals much longer than the packet interarrival time and, thus, allows reducing the amount of control traffic needed to configure the uplink grantfree transmissions.
In this paper, we assume that the gNB allocates dedicated channel resources to each UE, and transmission errors are only caused by fading. In our future works, we are going to consider a scenario in which several UEs can use shared grantfree resources to transmit their packets. We will adapt the proposed algorithm to take into account possible interference caused by the transmission of other UEs in shared resources.
References
 [1] “5G; Study on scenarios and requirements for next generation access technologies,” 3rd Generation Partnership Project (3GPP), Technical Report (TR) 38.913, 2020.
 [2] H. Ullah, N. G. Nair, A. Moore, C. Nugent, P. Muschamp, and M. Cuevas, “5G communication: an overview of vehicletoeverything, drones, and healthcare usecases,” IEEE Access, vol. 7, pp. 37 251–37 268, 2019.
 [3] “Framework and overall objectives of the future development of IMT for 2020 and beyond,” ITUR, Recommendation M.2083, September 2015.
 [4] E. Avdotin, D. Bankov, E. Khorov, and A. Lyakhov, “Enabling massive realtime applications in IEEE 802.11 be networks,” in Proceedings of the IEEE 30th Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC). IEEE, 2019, pp. 1–6.
 [5] “5G; NR; Physical layer procedures for data,” 3rd Generation Partnership Project (3GPP), Technical Specification (TS) 38.214, 2020.
 [6] D. Chase, “Code combining – a maximumlikelihood decoding approach for combining an arbitrary number of noisy packets,” IEEE transactions on communications, vol. 33, no. 5, pp. 385–393, 1985.
 [7] G. Berardinelli, N. Mahmood, R. Abreu, T. Jacobsen, K. Pedersen, I. Kovacs, and P. Mogensen, “Reliability analysis of uplink grantfree transmission over shared resources,” IEEE Access, vol. 6, pp. 23 602–23 611, 2018.
 [8] Y. Liu, Y. Deng, M. Elkashlan, A. Nallanathan, and G. Karagiannidis, “Analyzing grantfree access for URLLC service,” IEEE Journal on Selected Areas in Communications, 2020, early access.
 [9] N. Mahmood, R. Abreu, R. Bohnke, M. Schubert, G. Berardinelli, and T. Jacobsen, “Uplink grantfree access solutions for URLLC services in 5G new radio,” in Proceedings of the 16th International Symposium on Wireless Communication Systems (ISWCS). IEEE, 2019, pp. 607–612.
 [10] T. Jacobsen, R. Abreu, G. Berardinelli, K. Pedersen, P. Mogensen, I. Z. Kovacs, and T. Madsen, “System level analysis of uplink grantfree transmission for URLLC,” in Proceedings of the IEEE Globecom Workshops (GC Wkshps). IEEE, 2017, pp. 1–6.
 [11] M. C. LucasEstan, J. Gozalvez, and M. Sepulcre, “On the capacity of 5G NR grantfree scheduling with shared radio resources to support ultrareliable and lowlatency communications,” Sensors, vol. 19, no. 16, p. 3575, 2019.
 [12] B. Singh, O. Tirkkonen, Z. Li, and M. A. Uusitalo, “Contentionbased access for ultrareliable low latency uplink transmissions,” IEEE Wireless Communications Letters, vol. 7, no. 2, pp. 182–185, 2018.
 [13] T. Jacobsen, R. Abreu, G. Berardinelli, K. Pedersen, I. Z. Kovacs, and P. E. Mogensen, “System level analysis of Krepetition for uplink grantfree URLLC in 5G NR,” in Proceedings of the 25th European Wireless Conference. VDE, 2019, pp. 96–100.
 [14] Z. Zhou, R. Ratasuk, N. Mangalvedhe, and A. Ghosh, “Resource allocation for uplink grantfree ultrareliable and low latency communications,” in Proceedings of the IEEE 87th Vehicular Technology Conference (VTC Spring). IEEE, 2018, pp. 1–5.
 [15] S. Ozaku, Y. Shimbo, H. Suganuma, and F. Maehara, “Adaptive repetition control using terminal mobility for uplink grantfree URLLC,” in Proceedings of the IEEE 91st Vehicular Technology Conference (VTC Spring). IEEE, 2020, pp. 1–5.
 [16] I. F. Akyildiz, E. Khorov, A. Kiryanov, D. Kovkov, A. Krasilov, M. Liubogoshchev, D. Shmelkin, and S. Tang, “XStream: a new platform enabling communication between applications and the 5G network,” in Proceedings of the IEEE Globecom Workshops (GC Wkshps). IEEE, 2018, pp. 1–6.
 [17] S. Lagen, K. Wanuga, H. Elkotby, S. Goyal, N. Patriciello, and L. Giupponi, “New radio physical layer abstraction for systemlevel simulations of 5G networks,” in Proceedings of the IEEE International Conference on Communications (ICC). IEEE, 2020, pp. 1–7.
 [18] The ns3 network simulator. [Online]. Available: http://www.nsnam.org/
 [19] I. Parvez, A. Rahmati, I. Guvenc, A. I. Sarwat, and H. Dai, “A survey on low latency towards 5G: RAN, core network and caching solutions,” IEEE Communications Surveys & Tutorials, vol. 20, no. 4, pp. 3098–3130, 2018.
 [20] T. S. Rappaport, Wireless communications: principles and practice, 2nd ed. Prentice Hall, 2002.
 [21] “Base Station (BS) radio transmission and reception,” 3rd Generation Partnership Project (3GPP), Technical specification (TS) 36.104, 2019.
Comments
There are no comments yet.