Coexistence Mechanism between eMBB and uRLLC in 5G Wireless Networks

The wireless industries are going through different kinds of emerging applications and services, e.g., high-resolution video streaming, virtual reality (VR), augmented reality (AR), autonomous cars, smart cities and factories, smart grids, remote medical diagnosis, unmanned aerial vehicles (UAV), artificial intelligence (AI) based personal assistants, sensing, metering, monitoring etc, along with the explosive trends of mobile traffic [1]. It is foreseen that the mobile application market will flourish in a CAGR of $29.1\%$ during $2015-2020$ [2]. Energy efficiency, latency, reliability, data rate, etc are distinct for separate applications and services. To handle these diversified requirements, International Telecommunication Union (ITU) has already classified 5G services into uRLLC, mMTC, and eMBB categories[3]. Gigabit per second (Gbps) level data rates are required for eMBB users, whereas connection density and energy efficiency are the major concern for mMTC, and uRLLC traffic focuses on extremely high reliability ($99.999\%$) and remarkably low latency ($0.25\sim 0.30$ ms/packet)[4].

Generally, the lions’ share of wireless traffic is produced by eMBB UEs. uRLLC traffic is naturally infrequent and needs to be addressed spontaneously. The easiest way to settle this matter is to allocate some resources for uRLLC. However, under-utilization of radio resources may emerge from this approach, and generally, effective multiplexing of traffics is required. For efficient multiplexing of eMBB and uRLLC traffics, 3GPP has recommended a superposition/puncturing skeleton [4] and the short-TTI/puncturing approaches [5] in 5G cellular systems. Though the short-TTI mechanism is straightforward for implementation, it degrades spectral efficiency because of the massive overhead in the control channel. On the contrary, the puncturing strategy decreases the above overhead, although it necessitates an adequate mechanism for recognizing and healing the punctured case. Slot ($1$ ms) and mini-slot ($0.125$ ms) are proposed as time units for meeting the latency requirement of uRLLC traffic in the 5G NR. At the outset of a slot, eMBB traffic is scheduled and continues unchanged throughout the slot. If the same physical resources are used, uRLLC traffic is overridden upon the scheduled eMBB transmission.

Currently, much attention has been paid to resource sharing for offering QoS or QoE to the users. Studies [6] and [7] investigate the sharing of an unlicensed spectrum between LTE and WiFi networks, however, the study [8] con sider LTE-A and NB-IoT services for sharing the same resources. Study [9] solves user association and resource allocation problems. The study [9] consider the downlink of fog network to support QoS provisions of the uRLLC and eMBB. Some other studies, however, investigates and/or analyzes the influence of uRLLC traffic on eMBB [10, 11, 12, 13, 14, 15] or presents architecture and/or framework for co-scheduling of eMBB and uRLLC traffic [16, 17, 18, 19]. Moreover, some authors consider eMBB and uRLLC traffic in their coexisting/multiplexing proposals [20, 21, 22, 23, 24, 25, 26, 27] where they apply puncturing technique.

As per our knowledge, concrete mathematical models and solutions, however, are lacking in most of these coexistence mechanisms. Most of the studies mainly focus on analysis, system-level design or framework. Thus, effective coexistence proposals between eMBB and uRLLC traffic are wanting in literature. So, to enable eMBB and uRLLC services in $5$G wireless networks, we propose an effective coexistence mechanism in this paper. Our preliminary work has been published in [24] where we have used a one-sided matching and heuristic algorithm, respectively, for resolving resource allocation problems of eMBB and uRLLC users. The major difference between [24] and current work is the involvement of PSUM and TM for solving similar problems. This paper mainly focuses on the followings:

• •

  First, we formulate an optimization problem for eMBB UEs with some constraints, where the objective is to maximize the minimum expected rate of eMBB UEs over time.

• •

  Second, to solve the optimization problem effectively, we decompose it into two sub-problems: resource scheduling for eMBB UEs, and resource scheduling of uRLLC UEs. PSUM is used to solve the first sub-problem, whereas the TM is employed to solve the second one.

• •

  Third, we redefine the first sub-problem into a minimization problem for each slot and provide an algorithm based upon PSUM to obtain near-optimal solutions.

• •

  Fourth, we redefine the second sub-problem as a minimization problem for each mini-slot within every slot and present the algorithm based upon MCC and MODI methods of the transportation model to find an optimal solution of the second sub-problem.

• •

  Fifth, we also present a cost-effective heuristic algorithm for resolving the first sub-problem.

• •

  Finally, we perform a comprehensive experimental analysis for the proposed scheduling approach and compare the results, MEAR and fairness[41] of the eMBB UEs, with the PS [22], MUPS[26], RS, EDS, and MBS approaches.

The remainder of the paper is systematized as follows. In Section II, we present the literature review. We explain the system model and present the problem formulation in Section III. The proposed solution approach of the above-mentioned problem is addressed in Section IV. In Section V, we provide experimental investigation, discussion, and comparison concerning the proposed solution. Finally, we conclude the paper in Section VI.

Recently, both industry and academia focus on the study of multiplexing between eMBB traffic and uRLLC traffic on the same physical resources. Information-theoretic arguments-based performance analysis for eMBB and uRLLC traffic has performed in [10]. The authors consider both OMA and NOMA for uplink in C-RAN framework. An insight into the performance trade-offs among the eMBB and uRLLC traffic is explained in [10]. In [11], authors have introduced eMBB influenced minimization problem to protect the uRLLC traffic from the dominant eMBB services. This paper explores their proposal for the mobile front-haul environment. In [12], the authors present an effective solution for multiplexing different traffics on a shared resource. Particularly, they propose an effective radio resource distribution method between the uRLLC and eMBB service classes following trade-offs among the reliability, latency and spectral efficiency. Moreover, they investigate the uRLLC and eMBB performance adopting different conditions.

In order to 5G service provisioning (i.e., eMBB, mMTC and uRLLC services), the authors of [13] have studied radio resources slicing mechanism, where the performance of both orthogonal and non-orthogonal are analyzed. They have proposed a communication-theoretic model by considering the heterogeneity of 5G services. They also found that the non-orthogonal slicing is significantly better to perform instead of orthogonal slicing for those 5G service multiplexing. Recently, for 5G NR physical layer challenges and solution mechanisms of uRLLC traffic communications has been presented in [14], where they pay attention to the structure of packet and frame. Additionally, they focus on the improvement of scheduling and reliability mechanism for uRLLC traffic communication such that the coexistence of uRLLC with eMBB is established. In [15], the authors have been analyzed the designing principle of the 5G wireless network by employing low-latency and high-reliability for uRLLC traffic. To do this, they consider varying requirements of uRLLC services such as variation of delay, packet size, and reliability. To an extent, they explore different topology network architecture under the uncertainty.

The authors of [16, 17, 18] present a resilient frame formation for multiplexing the provisions of different users. In [16], the authors jointly MBB and mission-critical communication traffic by engaging dynamic TDD and TTI. In [17], the authors represent tractable multiplexing of MBB, MCC, and mMTC considering dynamic TTI. The authors of [18] present a holistic overview of the agile scheduling for 5G that incorporates multiple users. They envision an E2E QoS architecture to offer improved opportunities for application-layer scheduling functionality that ensures QoE for each user. $M/D/m/m$ queueing model-based system-level design has proposed for fulfilling uRLLC traffic demand in [19], where they exhibit that the static bandwidth partitioning is inefficient for eMBB and uRLLC traffic. Thus, the authors of [19] have illustrated a dynamic mechanism for multiplexing of eMBB and uRLLC traffic and apply this in both frequency and time domain.

The efficient way of network resource sharing for the eMBB and uRLLC is studied in [20] and [21]. A dynamic puncturing mechanism is proposed for uRLLC traffic in [20] within eMBB resources to increase the overall resource utilization in the network. To enhance the performance for decoding of eMBB traffic, a joint signal space diversity and dynamic puncturing schemes have proposed, where they improve the performance of component interleaving as well as rotation modulation. In [21], a joint scheduling problem is formulated for eMBB and uRLLC traffic in the goal of maximizing eMBB users’utility while satisfying stochastic demand for the uRLLC UEs. Specifically, they measure the loss of eMBB users for superposition/puncturing by introducing three models, which include linear, convex and threshold-based schemes. For reducing the queuing delay of the uRLLC traffic, the authors introduce punctured scheduling (PS) in [22]. In case of insufficient radio resource availability, the scheduler promptly overwrites a portion of the eMBB transmission by the uRLLC traffic. The scheduler improves the uRLLC latency performance; however, the performance of the eMBB users are profoundly deteriorated. The authors of [23] and [24] manifest the coexistence technique for enabling 5G wireless services like eMBB and uRLLC based upon a punctured scheme. The authors present an enhanced PS (EPS) scheduler to enable an improved ergodic capacity of the eMBB users in [25]. EPS is capable of recovering the lost information due to puncturing and partially. eMBB users are supposed to be cognizant about the corresponding resource that is being penetrated by uRLLC. Therefore, the victim eMBB users ignore the punctured resources from the erroneous chase condensing HARQ process. The authors of [26] propose a MUPS, where they discretize the trade-off among network system capacity and uRLLC performance. MUPS first tries to match the incoming uRLLC traffic inside an eMBB traffic in a conventional MU-MIMO transmission. MUPS serves the uRLLC traffic instantly by using PS if multi-user (MU) pairing cannot be entertained immediately. Though MUPS shows improved spectral efficiency, it is not feasible for uRLLC latency as MU pairing mostly depends on the rate maximization. Hence, the inter-user interference can further degrade the SINR quality of the uRLLC traffic, which can lead to reliability concerns. The authors of [27] propose a null-space-based preemptive scheduler (NSBPS) for jointly serving uRLLC and eMBB traffic in a densely populated 5G arrangement. The proposed approach ensures on-the-spot scheduling for the sporadic uRLLC traffic, while makes a minimal shock on the overall system outcome. The approach employs the system spatial degrees of freedom (SDoF) for uRLLC traffic for spontaneously providing a noise-free subspace. In [28], the authors present a risk-sensitive approach for allocating RBs to uRLLC traffic in the goal of minimizing the uncertainty of eMBB transmission. Particularly, they launch the Conditional Value at Risk (CVaR) for estimating the uncertainty of eMBB traffic in [28].

In this work, we consider a 5G network scenario with one gNB which supports a group of user equipment (UE) $\mathcal{E}$ requiring eMBB service, and a set of user equipment $\mathcal{U}$ demanding uRLLC service. The system operates in downlink mode for the UEs and the overall system diagram is shown in Fig. 1. gNB supports the UEs using licensed RBs $\mathcal{K}$ each with equal bandwidth of $B$. Every time slot, with a length $\Delta$, is split into $M$ mini-slots of duration $\delta$ for managing low latency services. For supporting eMBB UEs, we consider $T_{s}$ LTE time slots and denoted by $\mathcal{T}=\{1,2,\cdots,T_{s}\}$. uRLLC traffic arrive at gNB (any mini-slot $m$ of time slot $t$) follows Gaussian distribution, i.e., $U\sim\mathcal{N}(\mu,\sigma^{2})$. Here, $\mu$ and $\sigma^{2}$ denote the mean and variance of $U$. Each uRLLC UE $u\in\mathcal{U}$ request for a payload of size $L_{u}^{m,t}$ (varying from 32 to 200 Bytes [29]).

gNB allots the RBs to the eMBB UEs at the commencement of any time slot $t\in\mathcal{T}$. The achievable rate of $e\in\mathcal{E}$ for RB $k\in\mathcal{K}$ is as follows:

where $\gamma_{e,k}^{t}=\frac{P_{e}h_{e}^{2}}{N_{0}B}$ presents SNR. $P_{e}$ is the transmission power of gNB for $e\in\mathcal{E}$ and $h_{e}$ denotes the gain of $e\in\mathcal{E}$ from the gNB, and $N_{0}$ represnts the noise spectral density. eMBB UEs require more than one RB for satisfying their QoS. Therefore, the achievable rate of eMBB UE $e\in\mathcal{E}$ in time slot $t$ as follows:

where $\boldsymbol{\alpha}$ denotes the resource allocation vector for $\mathcal{E}$ at any time slot $t$, and each element is as follows:

uRLLC traffic can arrive at some moment (i.e. mini-slot) inside any time slot $t$ and requires to be attended quickly. Any uRLLC traffic needs to be completed within a mini-slot period for its’ latency and reliability constraints. Normally, the payload size of uRLLC traffic is really short, and therefore, we cannot straightforwardly adopt Shannon’s data rate formulation[10]. The achievable rate of a uRLLC UE $u\in\mathcal{U}$ in RB $k\in\mathcal{K}$, when its’ traffic is overlapped with eMBB traffic, can properly be approximated by employing [30] as follows:

where $\gamma_{u}^{m,t}=\frac{h_{u}^{2}P_{u}}{N_{0}B+h_{u}^{2}P_{e}}$ represents the SINR for $u\in\mathcal{U}$ at mini-slot $m$ of $t$. Here, $h_{u}^{2}P_{e}$ indicates the interference generated from serving $e\in\mathcal{E}$ in the same RB, $V_{u}=\frac{h_{u}^{2}P_{u}}{N_{0}B+h_{u}^{2}(P_{u}+P_{e})}$ depicts the channel dispersion, and meaning of other symbols are shown in II. However, the reliability of uRLLC traffic fall into vulnerability due to the interference. Hence, superposition mechanism is not a suitable for serving uRLLC UE[11]. Thus, for serving uRLLC UEs, we concentrate on the puncturing technique . In the punctured mini-slot, gNB allots zero power for eMBB UE, and therefore, the interference cannot affect the uRLLC traffic. At that time, $\gamma_{u}^{m,t}=\frac{h_{u}^{2}P_{u}}{N_{0}B}$ and $V=\frac{h_{u}^{2}P_{u}}{N_{0}B+h_{u}^{2}P_{u}}$. The achieved rate of $u\in\mathcal{U}$, when it uses multiple RBs, is as follows:

where $\boldsymbol{\beta}$ is the resource allocation vector for $\mathcal{U}$ at $m$ of $t$, and each of its’ element follows:

All the uRLLC request in any $m$ of $t$ needs to be served for sure, and hence,

where $\boldsymbol{\phi}$ denotes a vector for the serving uRLLC UEs, and thus,

Within the stipulated period $\delta$, the payload $L_{u}^{m,t}$ of $u\in\mathcal{U}$ needs to be transferred, and hence, satisfy the following:

Hence, the reliability and latency concerns of uRLLC traffic are simultaneously shielded by (7) and (9). Besides, $e\in\mathcal{E}$ loses some throughput at $t$ if uRLLC traffic is punctured within its’ RBs. We utilize the linear model of [21] for estimating the throughput-losses of eMBB UE. Therefore, the throughput-losses $e\in\mathcal{E}$ looks like as follows:

So, the actual achievable rate of $e\in\mathcal{E}$ in any $t$ is as follows:

We see that $\boldsymbol{\beta}$ affects on $\boldsymbol{\alpha}$, and hence, impact negatively to the eMBB throughput in each $t\in\mathcal{T}$. At the start of any $t\in\mathcal{T}$, gNB allocates the RBs $\mathcal{K}$ among the $\mathcal{E}$ in an orthogonal fashion as shown in Fig. 2. These characteristics of $\boldsymbol{\alpha}$ are shown mathematically as follows:

Within each $t\in\mathcal{T}$, gNB allows uRLLC UEs to get some RBs immediately on a mini-slot basis. Therefore, uRLLC traffic overlaps with eMBB traffic at $m$ and also shown in Fig. 2. Accordingly, $\boldsymbol{\beta}$ satisfy the following conditions on each $m$:

Finally, our objective is to maximize the actual achievable rate of each eMBB UE across $\mathcal{T}$ while entertaining nearly every uRLLC request within its’ speculated latency. We apply Max-Min fairness doctrine for this mission, and it contributes stationary service quality, enhances spectral efficiency and makes UEs more pleasant in the network. Hence, the maximization problem is formulated as follows:

In (18), the reliability and latency constraints of the uRLLC UEs are preserved by (18a) and (18b). Constraints (18c) and (18d) are used to show the orthogonality of RBs among eMBB and uRLLC UEs, respectively. At least one RB is posed by every active UE and is encapsulated by both (18e) and (18f). Resource restriction is presented by constraint (18g). Constraint (18h) shows that every item of $\boldsymbol{\alpha}$, $\boldsymbol{\beta}$ and $\boldsymbol{\phi}$ are binary. The formulation (18) is a Combinatorial Programming (CP) problem having chance constraint, and NP-hard due to its nature.

We assume that eMBB UEs are data-hungry over the considered period. Thus, at the commencement of a time slot $t\in\mathcal{T}$, gNB schedules all of its’ RBs among the eMBB UEs and stay unchanged over $t$. If uRLLC traffic requests come in any $m$ of $t$, the scheduler tries to serve the requests in the next $m+1$. Hence, the overlapping of uRLLC traffic over eMBB traffic happens as shown in Fig. 2. Usually, a portion of all RBs is required for serving such uRLLC traffic. However, the challenge is to find the victimized eMBB UE(s) following the aspiration of the problem (18).

For getting an effective solution to the problem (18), we can utilize the concept of a divide-and-conquer strategy. Here, we divide (18) into two resource allocation sub-problems, namely, for eMBB UEs on time slot basis and uRLLC UEs on a mini-slot basis. The first sub-problem is as follows:

On the other hand, the second sub-problem (with $\alpha^{t},\forall t$ as the solution of 19) is manifested as follows:

Fig. 3 shows the solution overview of the optimization problem (18). We can better understand the philosophy of the problem and the solution approach with an illustrative example in Fig. 2. At the beginning of the time slot, $t-1$, let us assume that there are 3 eMBB UEs, each of whom owns 4 RBs. Within $t-1$, the service request for uRLLC UEs came abruptly and the allocation of RBs for that UEs is shown in Fig. 2, as overlapped uRLLC traffic in the mini-slots. During this time, eMBB users $1,2$ and $3$ waste throughput equivalent to $4$RBs$\times 1$ mini-slot, $7$RBs$\times 1$ mini-slot, and $2$RBs$\times 1$ mini-slot, respectively. At the start of the next time slot, $t$, gNB acknowledges the resource scheduling of uRLLC UEs of $t-1$ to allocate and compensate eMBB UEs. gNB allocates more RBs to eMBB user 2 and less to eMBB user 3 as they lose more and less, respectively, in the time slot $t-1$. Moreover, EgNB tries to serve uRLLC users such that the loss of throughput of eMBB users are almost similar in the time slot $t$. Therefore, gNB makes a balance among the throughput of eMBB users in each time slot, which ultimately serves to reach the goal of (18) on a long-run basis.

In this section, we assess the proposed approach using comprehensive experimental analyses. Here, we compare our results with the results of the following state-of-the-art schedulers:

• •

  PS[22]: PS immediately overwrite part of the continuing eMBB transmission with the sporadic uRLLC traffic if there are not sufficient PRBs available. It chooses PRBs with the highest MCS that already been allotted to eMBB UEs.

• •

  MUPS[26]: In case of insufficient RBs, MUPS allocates PRBs to the uRLLC UEs where they endure better channel quality depending on the CQI feedback.

• •

  RS: RS takes the RBs from the eMBB UEs randomly in case of inadequate PRBs for supporting uRLLC traffic.

• •

  EDS: For supporting sporadic uRLLC traffic, EDS offers the PRBs to this traffic after preempting PRBs equally from the eMBB UEs in case of unavailable PRBs.

• •

  MBS: gNB uses many to one matching game for snatching PRBs from eMBB UEs for supporting uRLLC traffic.

The main performance parameters are MEAR and fairness [41] of the eMBB UEs and defined as follows:

In our scenario, we consider an area with a radius of $200$ m and gNB resides in the middle of the considered area. eMBB and uRLLC UEs are disseminated randomly in the coverage space. gNB works on a $10$ MHz licensed band for supporting the UEs in downlink mode. Every uRLLC UE needs a single PRB for its service. Furthermore, gNB estimates path-loss for both eMBB and uRLLC UEs using a free space propagation model amidst Rayleigh fading. Table III exhibits the significant parameters for this experiment. We use similar PSUM parameters as of [31]. We realize the results of every approaches after taking $1,000$ runs.

A comparison of MEAR and fairness scores are presented in Fig. 4 and Fig. 5, respectively, between the proposed (PSUM+TM) and the optimal value for a small network. Fig. 4 shows the ECDF of MEAR and the probability of MEAR being at least $20$ Mbps are around $0.50$ and $0.70$, respectively, for the proposed and optimal methods, consequently. The optimality gap of average MEAR for the proposed method is $4.20\%$ as represented in Fig. 4. Fig. 5 shows the ECDF of the fairness scores where the probability of the scores being $0.995$ at least is $0.80$ in the proposed method in comparison of being $1$ in the optimal mechanism. The optimality gap of the proposed method for the average fairness score is $0.32\%$ as exposed from Fig. 5.