Baseband Functions Research Articles

Slice aware baseband function splitting and placement in disaggregated 5G Radio Access Network

Efficient utilization of network resources is becoming increasingly challenging for mobile network operators (MNOs) due to the widespread adoption of 5G and Beyond 5G (B5G) networks. With the introduction of functional splits in 5G Radio Access Network (RAN), the baseband functions are disaggregated and placed in geographically separated locations, remarkably improving the flexibility and efficiency of RAN. However, such disaggregation also brings new challenges when it comes to determining the most appropriate placement option for the baseband functions. Different functional splits and network slices have different requirements in terms of delay and data rate. Moreover, the processing nodes of the edge clouds and regional clouds, as well as the transport links, have different capacities. All these factors can affect the placement option for the baseband functions. Though significant efforts have been put into addressing these challenges in recent works, further exploration is required to develop more efficient baseband function placement solutions. To this end, we aim to develop a resource-efficient strategy that considers both functional splits and network slice-specific requirements, limited capacities of transport links, and processing nodes to reduce the Operational Expenditure (OPEX) of RAN. One of the significant factors contributing to the OPEX stems from the active processing nodes in the network. Hence, the primary goal of this work is to minimize the cost of active processing nodes in the edge and regional cloud. We propose an optimization model using Mixed Integer Linear Programming (MILP) that selects appropriate functional split, baseband function placement options, and the set of paths for routing the traffic from different RAN slices. We perform extensive simulations and show that the proposed model incurs a lower cost than baseline strategies while placing the baseband functions in the network. To address the high computational complexity of the optimization model, we also introduce a novel polynomial time heuristic algorithm. Although the heuristic incurs a higher cost than the MILP, we show that it takes significantly less execution time, making it applicable to large-scale deployment scenarios.

Black-box optimization for anticipated baseband-function placement in 5G networks

In the context of the ever-evolving 5G landscape, where network management and control are paramount, a new Radio Access Network (RAN) as emerged. This innovative RAN offers a revolutionary approach by enabling the flexible distribution of baseband functions across various nodes, all tailored to meet the ever-shifting demands of both system requirements and user traffic patterns. As users move within the network, the need to anticipate and strategically position these baseband functions becomes crucial for seamless network operation. Traditionally, this challenge has been tackled through a two-step process: first, forecasting traffic patterns, and then optimizing resource allocation accordingly. However, this approach falls short in guaranteeing an efficient placement when actual traffic demands surge onto the network. It often leads to resource overbooking, constraint violations, and excessive power consumption, putting strain on the network’s capabilities. In this paper, we introduce a novel framework based on a black-box optimization approach. This tool empowers prediction algorithms not just with historical traffic data but also with insights from optimization outcomes. The goal is to minimize a loss function related to power consumption and constraint violation: this ensures a predicted placement that is feasible and whose power is close to optimal. This approach ensures that the predicted placement is both feasible and power-efficient, bridging the gap between theoretical prediction and practical implementation. Remarkably, our proposed method, while potentially sacrificing some degree of traffic prediction accuracy, outperforms the conventional two-step approach by delivering a more efficient baseband function placement.

DRL-Based Energy-Efficient Baseband Function Deployments for Service-Oriented Open RAN

DRL-Based Energy-Efficient Baseband Function Deployments for Service-Oriented Open RAN

A low power circuit design of BLE baseband transmission data processing module

With the quickly development of the Internet of Things technology, Bluetooth (BLE) protocol has been increasingly adopted in consumer communication applications, industry communication utilization, and even positioning requirement, due to its advantages such as low power consumption, low cost, and low complexity. Especially the outstanding low power capability compared with other short-range communication protocol, many power limited AIoT (Artificial Intelligence IoT) devices select BLE as basic component. So, the low power, high performance, and small area BLE module become the key point of successful of whole AIoT system. In BLE system, the baseband data processing is the foundation. A good hardware circuit design of BLE baseband can improve BLE module performance and decrease power simultaneously. This paper analyzes the baseband data processing function of BLE in detail, and designs the baseband transmission data processing RTL circuit IP using Verilog HDL Hardware description language based on BLE protocol. The IP includes CRC verification module, whitening module and coding mapping function module. The RTL design is simulated and evaluated in workstation server with Synopsys VCS tool. The simulation results show that this design can achieve the Low-power BLE Bluetooth baseband function.

Minimizing energy consumption by joint radio and computing resource allocation in Cloud-RAN

Cloud-RAN is a key 5G enabler; it centralizes the baseband processing of several base stations by executing the baseband functions in a centralized, virtualized, and shared entity known as the Base Band Unit (BBU)-Pool. Cloud-RAN paves the way for joint management of the radio and computing resources of multiple base stations. In fact, centralization and virtualization allow for decreasing energy consumption which decreases Capital Expenditure (CAPEX) and Operational Expenditure (OPEX). Cloud-RAN architecture permits jointly allocating the radio and computing resources of multiple base stations. The radio resources include the Resource Blocks (RBs), the transmission power, and the Modulation Coding Scheme (MCS), whereas the computing resources include the CPUs resources. This paper investigates the potential benefits that could be scored thanks to the joint allocation of these two types of resources, with respect to energy consumption and overall throughput, when radio resources are finite and computing resources are not. The latter is an effect of the C-RAN architecture, which allows scalability and fast computing resource provisioning. Due to the unconstrained availability of computing resources, the joint allocation of radio and computing resources has a negligible impact when the objective is throughput maximization. However, it is highly beneficial when the target is energy consumption minimization in comparison to the sequential allocation that consists of allocating radio resources first, and then computing resources are allocated. For that, we formulate a Mixed Integer Linear Programming (MILP) problem having the objective of minimizing energy consumption. When the goal is to minimize energy consumption, the joint allocation of radio and computing resources reduces the total energy consumption by up to 21.3% when compared to the case where radio and computing resources in the BBU pool are allocated sequentially. Furthermore, given the NP-hardness of solving a MILP problem, we propose a two-step low-complexity matching game-based algorithm with a transmission power adjustment mechanism that aims at performing close to the MILP solver. The results show that our proposed matching game algorithm is a good alternative for solving the joint-allocation MILP problem, producing results that are very close to the MILP optimal solutions.

Fairness Guaranteed and Auction-Based x-Haul and Cloud Resource Allocation in Multi-Tenant O-RANs

The open-radio access network (O-RAN) embraces cloudification and network function virtualization for base-band function processing by dis-aggregated radio units (RUs), distributed units (DUs), and centralized units (CUs). These enable the cloud-RAN vision in full, where multiple mobile network operators (MNOs) can install their proprietary or open RUs, but lease on-demand computational resources for DU-CU functions from commonly available open-clouds via open x-haul interfaces. In this paper, we propose and compare the performances of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">min-max fairness</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Vickrey-Clarke-Groves (VCG) auction</i> -based x-haul and DU-CU resource allocation mechanisms to create a multi-tenant O-RAN ecosystem that is sustainable for small, medium, and large MNOs. The min-max fair approach <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">minimizes the maximum OPEX of RUs</i> through cost-sharing proportional to their demands, whereas the VCG auction-based approach <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">minimizes the total OPEX for all resources utilized while extracting truthful demands from RUs</i> . We consider time-wavelength division multiplexed (TWDM) passive optical network (PON)-based x-haul interfaces where PON virtualization technique is used to flexibly provide optical connections among RUs and edge-clouds at macro-cell RU locations as well as open-clouds at the central office locations. Moreover, we design efficient heuristics that yield significantly better economic efficiency and network resource utilization than conventional greedy resource allocation algorithms and reinforcement learning-based algorithms.

Slice Aware Baseband Function Placement in 5G RAN Using Functional and Traffic Split

5G and Beyond 5G (B5G) are undergoing numerous architectural changes to enable higher flexibility and efficiency in mobile networks. Unlike traditional mobile networks, baseband functions in 5G are disaggregated into multiple components - Radio Unit (RU), Distributed Unit (DU), and Centralized Unit (CU). These components can be placed in different geographical locations based on the latency sensitivity and available capacity in the network. Processing baseband functions in a centralized location offer various advantages (known as centralization benefit in RAN) to mobile network operators, which have been a point of interest for several research works. However, achieving maximum centralization is challenging due to various factors such as limited capacity in the midhaul network, delay requirement of different functional splits and network slices, etc. In this work, we aim to address these challenges by proposing a slice-aware baseband function placement strategy. Our primary objective is to maximize the degree of centralization in the network by appropriate selection of functional split. To achieve this objective, we jointly consider functional split, traffic split, different placement options for baseband functions, and network slice-specific requirements. We also consider the minimization of active processing nodes in cloud infrastructure of different levels (edge and regional) to provide additional resource efficiency. To this end, we formulate an optimization model using Mixed Integer Linear Programming (MILP) and compare its performance with different baseline techniques. We show that the proposed model achieves 6.5% more degree of centralization than the state-of-the-art while placing baseband functions in the network. To tackle the high computational complexity of the MILP model, we also present a polynomial-time heuristic algorithm for solving the problem in large-scale scenarios. We show that although the optimization model achieves around 4% more degree of centralization than the heuristic, the heuristic solves the problem in a reasonable amount of time, making it suitable for real deployment scenarios.

Baseband-Function Placement With Multi-Task Traffic Prediction for 5G Radio Access Networks

The 5G Radio Access Network (RAN) virtualization aims to improve network quality and lower the operator’s costs. One of its main features is the functional split, i.e., dividing the instantiation of RAN baseband functions into different units over metro-network nodes. However, its optimal placement is non-trivial: it depends on the application requirements and on the expected traffic volume, whose daily variation highly impacts the total power consumption. Current optimization solutions fail to provide a placement solution capable of handling traffic fluctuations. In fact, the standard machine learning algorithms used in the literature for planning the network resources in advance result in an allocation that is inadequate to carry the actual traffic at all the time-slots. Hence, we must reserve an artificial buffer capacity in the nodes to ensure feasibility. Instead, our proposed method exploits a fine-grained two-step multi-task algorithm that predicts the mean and quantile traffic, making the artificial capacity no longer necessary. The subsequent placement uses mixed-integer linear programming and a heuristic. The former considers the expected traffic in the objective function (to estimate costs) and the quantile in the constraints (to enforce capacity limits). The heuristic combines the mean and quantile results to minimize the power and comply with the requirements. While using sufficiently large artificial buffers guarantees robustness with a mild power increase compared to the oracle, the fine-grained multi-task model improves the results, reducing the power consumption compared to the mean and meets all constraints. The heuristic enables significant computational time reduction.

Edge-enhanced graph neural network for DU-CU placement and lightpath provision in X-Haul networks

The emerging mobile services have imposed several new challenges on the radio access network (RAN), which stimulates its evolution from cloud-RAN to next-generation RAN (NG-RAN). In NG-RAN, baseband functions are redefined as the radio unit (RU), distributed unit (DU), and central unit (CU), which are, respectively, connected by optical front/mid/backhaul (X-Haul) networks. The DU-CU placement and lightpath provision need an elaborate strategy for allocating resources tailored to user requests. However, most existing methods fail to fully perceive network conditions and then make inappropriate solutions, which may result in over-consumption of processing and transmission resources. Therefore, we propose a reinforcement-learning-based DU-CU placement and lightpath provision strategy using an edge-enhanced graph neural network, i.e, EGNN-RL. The EGNN is leveraged to adequately exploit graph-structured features in X-Haul networks, while proximal-policy-optimization-based RL is introduced to maintain policy stability. In addition, this problem is formulated as an integer linear programming model to find the optimal solution. We validate the proposed strategy under different service types (i.e., enhanced mobile broadband, ultra-reliable low latency communication, and massive machine connections) and different numbers of services, respectively. The results show that our scheme can achieve higher resource efficiency compared with existing methods. Moreover, it also adapts to new networks with the same nodes through fine-tuning the model. This can decrease nearly 15% of the retaining time and obtain similar performance with retaining.

Power-Efficient Baseband-Function Placement in Latency-Constrained 5G Metro Access

Mobile operators face the challenge of deploying a flexible network to handle the requirements of emerging use cases. A new Radio Access Network (RAN) for 5G was proposed to provide more dynamic and energy-efficient network management, resource allocation, and service provisioning compared to 4G. The baseband functions, previously performed next to the antennas, are divided into different radio-network units – Radio Unit (RU), Distributed Unit (DU), Centralized Unit (CU) – according to functional splits, which define how to distribute the functions in the various units. The operators encounter the problem of placing these units over the network to minimize their cost and guarantee Quality of Service (QoS). In this work, we propose a Mixed Integer Linear Programming (MILP) formulation that minimizes the DU/CU placement power consumption subject to functional split, latency, and capacity constraints. We also develop a heuristic algorithm with reduced complexity for larger topologies. The results show that optimizing the DU/CU placement using either the MILP or the heuristic algorithms saves power compared to the fully decentralized scenario and guarantees compliance to all requirements. Moreover, despite not achieving the exact MILP result, we show that the heuristic placement reaches a similar power consumption and respects all constraints.

Evolution Trends and Paradigms of Low Noise Frequency Synthesis and Signal Conversion Using Silicon Technologies

Silicon technologies for HF applications have been proven for more than two decades, and technologies have greatly evolved. Whether CMOS or BiCMOS technologies, the unique combination of radio frequency, baseband, and digital functions allow a very high level of integration. While it is possible to achieve fully integrated transceivers, the major advantages of these silicon technologies lie mainly in their unparalleled performance in the field of frequency synthesis and frequency conversion. We propose in this paper a review of the major results obtained on these RF components since the beginning of the 2000s, also considering the impact of the technology node. The back-end of line (BEOL) process on which depends the quality of microwave monolithic integrated circuits (MMICs) is briefly presented in the introductory part. If circuit performances are tightly bound to the active devices (i.e., the heterojunction bipolar transistor SiGe HBT or CMOS transistor), passive elements (i.e., quality factor of inductors and varactors, losses of transmission, or interconnection lines) as well as the definition of the substrate also play a major role. The core of the article is oriented toward the noise of synthesized signals and frequency conversion. Frequency synthesis is presented through the analog design of a voltage-controlled oscillator (VCO) or through the direct digital frequency synthesis (DDFS), for which respective figures of merit are presented and discussed in a second section. The spectral purity of the oscillators being decisive in the definition of the throughput of a link is approached through the comparison of different figures of merit (FoM) for a set of circuits achievements over the selected period. If the realization of free oscillators is closely bound to the phase-locked loop (PLL)-type control loop for VCOs, the DDFS solution provides more direct and more flexible alternative at first sight. Therefore, these two solutions are analyzed collectively. Finally, the oscillator integrated in the transmitter or receiver supplies the needed LO (local oscillator) power to the frequency mixer in the frequency conversion module: henceforth, the third part of this study focuses on high-frequency mixer realizations. We thus consider this LO power in some advanced figure of merit mentioned in the second section. The design trade-off of the mixer is presented in an approach combining LO (conversion gain, channel isolation, and phase noise) and RF (HF noise figure and channel isolation) constraints. The final section provides a summary of the results and trends mentioned in the paper.

Deep Reinforcement Learning-Based Policy for Baseband Function Placement and Routing of RAN in 5G and Beyond

In this paper, we propose a deep reinforcement learning (DRL)-based algorithm to generate policies of Baseband Function (BBF) placement and routing. In order to explore the performance of the proposed algorithm in practical systems, the online scenario with the completely random requests is used in the simulation considering C-RAN and NG-RAN architectures. Besides, an Integer Linear Programming (ILP) model is formulated to generate the optimal solution as the benchmark. The simulation results show that DRL-based algorithm converges in a short time, and its performance closes to the optimal benchmark obtained by ILP in terms of latency and bandwidth for the online scenarios. In addition, the performance of the generated policies based on DRL is compared with a classic heuristic algorithm, i.e., first-fit algorithm. The performance of DRL-based algorithm is superior to the first-fit algorithm from above two perspectives. The fast convergence and the near-optimal performance prove that the DRL-based algorithm is a promising approach for the BBF placement and routing of RAN in 5G and Beyond.

Acoustic time-dependent energy from vibrating surfaces via a generalized radiation impulse response approach.

A vibrating surface in contact with a fluid experiences a reaction force that is related to the velocity of the surface. In the case of a harmonic vibration, the acoustic radiation impedance provides a useful measure of the harmonic force resulting from a harmonic velocity of the surface. More generally, the radiation impulse response, which is the temporal Fourier transform of the radiation impedance, provides the basis for a convolution approach to evaluate both the time-dependent force and the energy transfer into the fluid resulting from a time-dependent velocity of a fluid-loaded vibrating surface for interior and external problems. Important properties of the radiation impulse responses are presented using the Kirchhoff Surface Integral Equation and Fourier transforms. A universal form of the radiation impulse response, which consists of a weighted Dirac delta function and a baseband function, is presented. The time-dependent force and energy transfer into a fluid resulting from the space-time normal velocity of a spherical surface is presented to simply illustrate the generalized radiation impulse response approach for exterior three-dimensional radiation problems.

Uplink Latency in Massive MIMO-Based C-RAN With Intra-PHY Functional Split

Functional splitting and packetized fronthaul (FH) are two approaches to realize cloud radio access networks in a cost-effective manner. In the former, some baseband functionalities are offloaded to remote radio units (RRUs) instead of centralizing all of them at a baseband unit pool. This reduces the capacity and latency requirements on FH when massive multiple-input-massive-output RRUs are used. The latter approach aims to use the ubiquitous Ethernet networks for FH. However, this leads to random packet delays due to queuing at switching/aggregating gateways. In this letter, we present a novel analytical framework to characterize the distribution of queuing delays at an aggregation gateway in the uplink. This framework incorporates random user activity, the uplink spectral efficiencies of users, the slotted nature of uplink transmissions, and FH capacity. We study the impact of packet arrival rates, average packet sizes, and FH capacity on queue length and queuing delay distributions. We show that significant statistical multiplexing gains are possible by aggregating traffic from multiple RRUs.

Scalable Edge Computing Deployment for Reliable Service Provisioning in Vehicular Networks

The global connected cars market is growing rapidly. Novel services will be offered to vehicles, many of them requiring low-latency and high-reliability networking solutions. The Cloud Radio Access Network (C-RAN) paradigm, thanks to the centralization and virtualization of baseband functions, offers numerous advantages in terms of costs and mobile radio performance. C-RAN can be deployed in conjunction with a Multi-access Edge Computing (MEC) infrastructure, bringing services close to vehicles supporting time-critical applications. However, a massive deployment of computational resources at the edge may be costly, especially when reliability requirements demand deployment of redundant resources. In this context, cost optimization based on integer linear programming may result in being too complex when the number of involved nodes is more than a few tens. This paper proposes a scalable approach for C-RAN and MEC computational resource deployment with protection against single-edge node failure. A two-step hybrid model is proposed to alleviate the computational complexity of the integer programming model when edge computing resources are located in physical nodes. Results show the effectiveness of the proposed hybrid strategy in finding optimal or near-optimal solutions with different network sizes and with affordable computational effort.

Flexible fine-grained baseband processing with network functions virtualization: Benefits and impacts

The increasing demand for wireless broadband connectivity is leading mobile network operators towards new means to expand their infrastructures efficiently and without increasing the cost of operation. Network Functions Virtualization (NFV) is a step towards virtualization-based, low-cost flexible and adaptable networking services. In the context of centralized baseband architectures, virtualization is already employed to run baseband processing units as software on top of conventional data center hardware. However, current virtualization solutions consider atomic virtualization, i.e., single virtual machines implementing all baseband functionalities. In this article, we propose the fine-grained virtualization of baseband processing to achieve a more flexible distribution of the processing workload in centralized architectures. We also evaluate the benefits of our approach in terms of: (i) The bandwidth requirements for each fine-grained distribution option, (ii) the latency experienced by mobile users for each fine-grained distribution option, and (iii) the total CPU usage of each fine-grained baseband processing function.

Centralized vs. distributed algorithms for resilient 5G access networks

Cloud radio access networks (C-RANs), relying on network function virtualization and software-defined networking (SDN), require a proper placement of baseband functionalities (BBUs) to reach full coverage of served areas and service continuity. In this context, network resources can be shared and orchestrated to meet the flexibility required by a dynamically evolving environment. Different methodologies, based on analytical formulation or heuristic algorithms, can be applied to achieve suitable trade-offs among cost components. This paper considers both centralized and distributed algorithms to obtain BBU hotel placement in C-RAN and compares their performance, scalability and adaptability to evolving scenarios. As expected, the results obtained with the distributed approach are sub-optimal, but very close, in most cases, to the optimal solutions obtained with a centralized algorithm based on integer linear programming. In addition to off-loading the SDN orchestrator, the distributed approach, differently from the centralized one, is shown to be able to cope with the evolution of the C-RAN topology with limited incremental changes in the original placement. The limits of the centralized approach in terms of scalability that the distributed approach is able to overcome are also evidenced.

Optimized Energy Aware 5G Network Function Virtualization

In this paper, network function virtualization (NVF) is identified as a promising key technology that can contribute to energy-efficiency improvement in 5G networks. An optical network supported architecture is proposed and investigated in this work to provide the wired infrastructure needed in 5G networks and to support NFV towards an energy efficient 5G network. In this architecture the mobile core network functions as well as baseband function are virtualized and provided as VMs. The impact of the total number of active users in the network, backhaul/fronthaul configurations and VM inter-traffic are investigated. A mixed integer linear programming (MILP) optimization model is developed with the objective of minimizing the total power consumption by optimizing the VMs location and VMs servers’ utilization. The MILP model results show that virtualization can result in up to 38% (average 34%) energy saving. The results also reveal how the total number of active users affects the baseband virtual machines (BBUVMs) optimal distribution whilst the core network virtual machines (CNVMs) distribution is affected mainly by the inter-traffic between the VMs. For real-time implementation, two heuristics are developed, an Energy Efficient NFV without CNVMs inter-traffic (EENFVnoITr) heuristic and an Energy Efficient NFV with CNVMs inter-traffic (EENFVwithITr) heuristic, both produce comparable results to the optimal MILP results. Finally, a Genetic algorithm is developed for further verification of the results.

Tradeoff between User Quality-Of-Experience and Service Provider Profit in 5G Cloud Radio Access Network

In recent years, the Cloud Radio Access Network (CRAN) has become a promising solution for increasing network capacity in terms of high data rates and low latencies for fifth-generation (5G) cellular networks. In CRAN, the traditional base stations (BSs) are decoupled into remote radio heads (RRHs) and base band units (BBUs) that are respectively responsible for radio and baseband functionalities. The RRHs are geographically proximated whereas the the BBUs are pooled in a centralized cloud named BBU pool. This virtualized architecture facilitates the system to offer high computation and communication loads from the impetuous rise of mobile devices and applications. Heterogeneous service requests from the devices to different RRHs are now sent to the BBUs to process centrally. Meeting the baseband processing of heterogeneous requests while keeping their Quality-of-Service (QoS) requirements with the limited computational resources as well as enhancing service provider profit is a challenging multi-constraint problem. In this work, a multi-objective non-linear programming solution to the Quality-of-Experience (QoE) and Profit-aware Resource Allocation problem is developed which makes a trade-off in between the two. Two computationally viable scheduling algorithms, named First Fit Satisfaction and First Fit Profit algorithms, are developed to focus on maximization of user QoE and profit, respectively, while keeping the minimum requirement level for the other one. The simulation environment is built on a relevant simulation toolkit. The experimental results demonstrate that the proposed system outperforms state-of-the-art works well across the requests QoS, average waiting time, user QoE, and service provider profit.

Synergetic communication-and-computation optimization in software-defined hyper-cellular networks

Recently, the cloud radio access network (C-RAN) architecture has been proposed to enhance the cost effectiveness and flexibility of traditional cell-centric radio access networks. However, the massive fronthaul bandwidth required to centralize baseband computations in C-RAN results in extremely high costs. This paper summarizes our previous efforts toward solving this problem. We proposed the software-defined hyper-cellular network (SDHCN) based on the control/data separation principle. Under the proposed SDHCN framework, we studied two mechanisms that can greatly reduce fronthaul costs through the joint deployment of communicational and computational resources. First, we quantitatively characterized the relationship between the size of virtual base station (VBS) pools and the gains from computational statistical multiplexing by using queueing theory. We then showed that the marginal gain diminishes quickly with a growing pool size. Therefore, it is most economical to deploy mid-sized VBS pools. Finally, we proposed a genetic algorithm for baseband function splitting within a graph-clustering framework. This algorithm provides splitting schemes that can flexibly achieve different tradeoffs between fronthaul and computational costs based on different design preferences.