An overview of LTE/LTE-A heterogeneous networks for 5G and beyond

The potential of fifth-generation (5G) and beyond wireless connectivity is to deliver higher data rates, much lower latency, more frequent increase in channel capacity, and significant improvements in Quality of Service (QoS), compared to the current Long Term Evolution (LTE) networks. The continued growth of smart devices, the introduction of trending multimedia applications, and the dramatic increase in demand and use of wireless data communication are already causing a significant load on current mobile networks. 5G wireless networks, with enhanced user data rates, delays, capacity and QoS, are anticipated to solve many of the existing problems of mobile networks. In this study, we review the emergent wireless LTE-A and 5G networks to provide an overview of the challenges and the solutions suggested in recent literature towards forth-coming 6G networks. We first discuss the new structural changes associated with the Radio Access Network (RAN) design, heterogeneous network and LTE or LTE-Advanced (LTE-A) network. To understand the challenges and research gaps in 5G and beyond networks, the paper reviews the outstanding features of the new QoS and Self-Organization Networks (SON) related to the emerging 5G networks. Data rates, bandwidth and coverage have been given significance throughout this review because these are some major challenges towards 6G networks. Since understanding the present state of 5G usage is critical to its adoption, the paper also discusses related field tests, trials, and simulation tests. Finally, we identify existing major research problems and outline potential research areas.

TA B L E 1 Architectural differences of 1G, 2G, 3G, 4G, 5G, and 6G. Wideband Code-Division Multiple Access (WCDMA), and Multiple Input Multiple Output (MIMO) have contributed significantly to this gradual transformation. The growing use of smart devices and multimedia applications, such as video conferencing, video streaming, e-health care and online games, are not only meeting the needs of users but also opening new business opportunities for organizations and mobile network operators. Due to the ever-increasing demand for multimedia applications, existing mobile networks will no longer be able to meet the bandwidth and quality of service requirements. Therefore, to enhance the quality of service, Long Term Evolution (LTE) has been proposed by 3GPP mobile systems. Later, in Release 10, LTE-Advanced (LTE-A) is recommended as 4G programs and beyond what is intended to be achieved in International Mobile Telecommunications (IMT)-improved performance standards.
For key radio access networks, LTE has been completely redesigned as a fully packet-switched optimized system with the dedicated architecture of Internet Protocol (IP), as set out in the definition of 3GPP (2013b). Orthogonal Frequency Division Multi-Access (OFDMA) is the key enabler for LTE systems, where the channel spectrum is divided into smaller radio systems called Physical Resource Blocks (PRB) (3GPP, 2012a). OFDMA is able to resist selective frequency blurring and internal disruption of cells. Plus, the more affordable the research blocks, the better is the access to Code Division Multiple Access (CDMA) in mobile systems. However, inter-cell disruption can affect OFDMA systems, so an efficient cell group integration scheme is necessary.
Mobile integration is where the user equipment (UE) device is connected to a specific base station (BS) of a cell following certain conditions set up by the operator, which we refer to as "cell association" or "cell selection". Appropriate cell association of the system not only contributes to the improvement of system performance but also to load balancing, which is another major problem in current mobile systems. Load balancing is the misbalance of users among Macro e-node Base Station (MeNB) and small base stations because of the disparity in their backhaul capacity. Macro-base station being more powerful, associates more users than Small Base Stations (SBS). Hence, the MeNB gets overloaded, and its users suffer a lack of channels required to maintain the 5G or beyond standards. Whereas, the SBS sits idle, having fewer users and not being properly utilized for what it has been deployed for. In addition, the LTE system is presumed to bear a few real-time applications, for example, VoIP and voice services, imposing Quality of Service (QoS) barriers. 1 Dealing with load balances and interference has many challenges, such as sorting the appropriate cell association scheme, interference mitigation schemes and radio resource management. Individual studies have been conducted, but a combination of these challenges in one scheme has rarely been evaluated.
To address these challenges, a new cell distribution concept called Heterogeneous Network (HetNet) has been proposed by 3GPP. This includes Low Power Nodes (LPNs), for example, femto-cells, pico-cells and relays mounted on the cover of MeNB to handle portable traffic, especially in tropical areas. Figure 1 shows the differences between a regular homogeneous network and a heterogeneous network. 2 Small cells in recent times have attracted a lot of interest from researchers and industry for the following reasons 3 : • Enhanced indoor coverage: In the near future, the forecast is that about 70% of data calls and 50% of calls will come from home locations. Unfortunately, indoor areas often suffer the loss of high-rise buildings. Thus, the Macro Base Station (MBS) signal is weakened or lost in these cases, called coverage holes of the macro-cell. Transmission of small cells to internal areas will bring better service due to the proximity between small cells and internal users.
• Offload traffic: The macro cell user load congestion is reduced by deploying more small base stations and handing over some users to those SBSs. Therefore, the MBS load has been shared between smaller cells, which is why users get better service.
• Cost reduction: Since large-scale shipping requires high installation expenses and cautious planning, deploying smaller primary channels is considered cost-effective compared to macro-cell. The expense of deploying small cells can be minimized by setting up an internal broadband connection. The insertion of small femto-cell-like cells depends on the simple idea of connecting and playing fast.
• Reduced consumption of power: Powerful signal is obtained by users via small cell deployment, rather than from a very small sub-station like a "relay node". Therefore, a very high transfer capacity is not required.
• QoS satisfaction improvement: As a small number of mobile users are usually provided with a small cell, each user can receive additional resources, thus leading to the development of QoS.
However, as MeNBs are costly to replace SBSs and there are differences in their transmission capacities, the major challenges of heterogeneous networks will be load balance or fairness, interference mitigation and resource allocation.
To overcome these problems, both researchers and the industry are conducting extensive research, which needs due consideration and further analysis.
The paper is organized as follows: Methodology is discussed in Section 2. Next, LTE and LTE-A architecture, heterogeneous networks have been discussed in Section 3. Section 4 presents radio resource management. The issues and F I G U R E 1 (A) Conventional homogeneous network versus (B) heterogeneous network. challenges for heterogeneous networks to achieve 5G and beyond networks are described in Section 5. Most of the relevant user association techniques used to acquire load balancing, resource management and interference mitigation have been discussed in Section 6. Section 7 presents the technical approaches for load balancing. Finally, a conclusion is provided.

METHODOLOGY
In this literature review, articles are included through an online database search for load balancing, interference mitigation and cell association. The most current (last 10 years) and related research objective papers were selected, and along with them, the journal impact factor and most cited papers were also prioritized. The PRISMA diagram in Figure 2 depicts the way the literature was shortlisted for inclusion in this paper. There were 110 articles found with the objectives: load balancing, interference mitigation and cell association. Out of these 110 articles, 100 full-text articles were found to be suitable. Next, we filtered out the same types of methodology papers, and finally, 51 articles were included as references in this review paper. These 51 articles were thoroughly studied and critically analyzed in this paper. The motivation of this paper is to study and analyze the LTE-A and 5G heterogeneous networks for the forthcoming 6G networks and concentrate mainly on user association, load balancing and interference mitigation. This work can be used as a reference for relevant works and to inform guidelines about future endeavors in this field.

OUTLINE OF HETEROGENEOUS LTE AND LTE-A NETWORKS
The following section depicts an outline of the LTE and LTE-A framework and the key features of the heterogeneous network.
F I G U R E 2 PRISMA flow chart.

LTE and LTE-A architecture
The LTE and LTE-A network uses System Architecture Evolution (SAE), which is an improved system structure consisting of a Radio Access Network (RAN) and a core network, as shown in Figure 2. Radio access network is known to evolve from Universal Terrestrial Radio Access Network (UTRAN), which includes UE and macro-cell BS which is known to have evolved Node Base stations (eNBs) in LTE or LTE-A context. Communication between E-UTRAN and Evolved Packet Core (EPC) is established by S1 communication via sub-gateway S-GW and eNB. The X2 link was brought up to make direct communication between eNBs signing. The core network is called the Evolved Packet Core (EPC), consisting of S-GW, Packet Gateway (P-GW) and Mobility Management Entity (MME).
In the case of EPC, the sub-gateway acts as a local travel anchor point of inter macro cell delivery and transit between 3GPP, too, regarding the management of IP packet transfers between EPC and related UE. User navigation and communication management between other 3GPP technologies and LTE or LTE-A are hosted by MME. P-GW acts between EPC and other similar internet protocol networks. In the LTE mobile system, the portable network bandwidth of 1.4 MHz to 20 MHz is used, and up to 100 MHz of spectrum is accessible to LTE-A. 4,5 Channel bandwidth is divided into smaller orthogonal Physical Resource Blocks (PRBs), 3 which is the smallest unit of radio equipment. Considering the short start of the cycle, each PRB contains 12 consecutive lower carriers and seven orthogonal frequencies to distinguish the multiplication symbols, where each PRB takes a 180 kHz spectrum and holds a time interval of 0.5 ms. For LTE or LTE-A programs, the radio framework contains 10 sub-frames, each of which has two intermediate spaces of 0.5 ms. So, the small LTE/LTE-A frame has a period of 1 ms, one Transmission Time Interval (TTI) and LTE/LTE-A radio frame save 10 ms.

Heterogeneous networks
LTE / LTE-A networks with different small cells, such as femto-cells, pico-cells, micro-cells and relay nodes, are depicted in Figure 3. The black shades in Figure 3 show the user density. Small cells are generally deployed to fill the macro-cell dark spot coverings and to expand capacity in highly dense traffic areas. This results in saving the cost of sending additional eNBs, 6,7 a process which requires careful network planning and efficiency. 8 A. Femto-cells: Femto-cell, also referred to as Home evolved NodeB (HeNB), acts as a short-range transmission power base station, especially to be used in indoor areas. 9 Transmission capacity in femto-cells generally ranges from 10 to

F I G U R E 3 LTE and LTE-A framework.
100 mW, while the coverage usually varies from 10 to 30 m. 10 Ad hoc installation alone is a very attractive femto-cell feature. This feature allows mobile operators to secure recovery costs as femto-cell user load can be handled by subscribers' broadband communication links (for example, digital subscriptions line, 9 fiber optic and satellite links 11 ) to the main network. These backhaul links also aid a small amount of user load from compatible BS, thereby decreasing traffic congestion eNB. Femto-cells can be given to a mobile operator with licensed macro-cell bands or different spectrum bands. However, the study in Reference 9 does not consider user association with multiple BSs. Also, the handover of the user from one base station to another, along with the mobility of the User Equipment (UE) has not been taken into account. The authors in References 12 and 13 showed base stations could work in a variety of ways of reaching, that is, open access, closed access and mixed access methods. The closed access mode is a femto-cell access control method, where members are only registered to a closed subscriber group of femto-cells. Such access mode is appropriate in residential areas. Femto-cells within open-access that allow all users to do so access them. In mixed access mode, all users have permission to access femto-cells but the subscriber group is prioritized. B. Relay Nodes: Relay node has been introduced in 3GPP Release 11 as other technologies supported by LTE-A systems.
For Relay Nodes (RN), active eNB is called Donor eNB (DeNB). 2 RN can work internally as well on out-of-band relaying modes. In the case of in-band relaying, the relay access link and relay backhaul link share similar network company frequencies, while using different network company frequencies for backhaul and links to access out-of-band relaying. Relay node is divided into Type-1 and Type-2. 14 A Type-1 relay node is an in-band transmission channel with its own body cell identity (ID), which appears as a separate cell from UE. This type of RN is called Layer 3 relay, which includes Most DeNB RRM functions that support rewind. Two types of Type-1 RN are also described, which are Type-1a and Type-1b, where the first one operates in out-of-band mode and the latest works in in-band mode. On the other hand, the Type-2 transmission is a Layer 2 in-band channel that transmits that it has no visible cell ID and only supports a few RRMs jobs. The Multi-cast Transmission Single-Frequency Network (MTSFN) subframe, which is utilized for streaming and streaming apps, was created to eliminate disruptions from the RN transmitter on its own receiver. It is optimized for approval of backhaul transmission only between RN and compliance DeNB. For LTE frames, up to six sub-frames can be configured as subframes under MTSFN. 14

RADIO RESOURCE MANAGEMENT
Handover management, packet scheduling, link adaptation, radio admission control, and other RRM functions in a heterogeneous cellular system include the distribution of resource blocks among macro-cells and small cells. Figure 4 depicts the RRM methods for LTE and LTE-A networks.
A. Resource Block Distribution Between Macro-cells And Small Cells: Depending on whatever RRM method is acceptable, resource allocation between macro-cells and small cells may be required in heterogeneous networks. Other issues in classifying resources in a network-based transfer include resource allocation between backhaul link, access link, and direct link (eNB-UE), depending on whether in-band mode or out-of-band mode is active. Disruption, resource requirement, batch size, user number, and so forth are all things to consider when practising cellular distribution. The eNB is usually where services are distributed. B. Handover Management: The RRC layer is responsible for managing user mobility and delivery. The assisted network controls and the delivery process for LTE/LTE-A applications, that is, the choice are taken at the eNB, while the signal power ratings are collected from the EU. Furthermore, LTE/LTE-A systems only support the hard drive, where radio services are provided by the eNB provider prior to new radio services being assigned to the targeted eNB. 2 In addition, for UE to achieve a specified BS, the random access process is granted throughout the grant process. Random access protocols that are based on contention and do not vary from contention are supported for LTE or LTE-A applications. 2 HeNB and Type-1 RN can be used for access control and supply management. C. Packet Scheduling: One of the key roles of RRM in Media Access Control (MAC) is powerful packet editing, which is done regularly in LTE/LTE-A applications. D. Link Adaptation: Another key issue is getting acclimated to the connectivity. The MAC layer's RRM function is used to obtain the top user effect operation with a particular target block error rate. 15 Link operations such as flexibility, coding and transmission power control should be aligned. The UE assigned here has a greater channel quality because of the improved modulation and coding technology (as indicated by CQI). Transmission control is frequently used in conjunction with AMC to boost cell output. Furthermore, the level of distractions can be managed by using such capabilities. E. Radio Admission Control: In the radio service management business, you may find the company regulating radio access. The decision on a new application for the adoption of a radio manager must be accepted in Layout 3 of the LTE / LTE-A protocol stack. The decision is taken in accordance with QoS standards for asking radio managers and radio service available for the usage of radio services. 2 To welcome the user, the radio's available resources must be sufficient to meet all user-related radio managers' QoS requirements. It could also depend on whether the user is a new native user from the cell or an offloaded user from a neighboring cell. But the authors in Reference 2 do not attempt to give sufficient consideration to load variation among macro-cell and small cells.

ISSUES AND CHALLENGES FOR HETEROGENEOUS NETWORKS TO ACHIEVE 5G AND BEYOND STANDARDS
Although the transfer of users to small cells benefits heterogeneous networks in several ways, a few problems and technical issues arise about such features to achieve 5G and beyond network standards: radio resource utilization, interference mitigation, QoS management, fairness, and complexity. The challenges and solutions to achieving 6G technology are summarized in Table 2. A. Radio Resource Utilization: One of the main issues for LTE-A networks is obtaining focused down-links and fast up-link data speeds. Effective radio resource utilization is crucial in light of contemporary issues such as restricted spectrum availability and the widespread use of low-bandwidth radios. The data rate and spectrum enhancement are TA B L E 2 Challenges to reach 6G from present 4G and 5G technology.

Challenges Description Solutions
Enhanced network latency, reliability and user experience The challenge includes a full automated and intelligent transformation of the whole network system consisting of massive amounts of data and unlimited computing resources Application of artificial intelligence (AI) technique. And it should be deployed not only on a particular location or on a device, but on the entire network 100% coverage Still more than 3 billion people in the world are unable to access the internet. The reasons are difficulty of installation of BSs in remote areas, higher expense to deploy numerous BS and there are also limitations of geographical conditions It has become a consensus of the industry to build an air-space-earth integrated 3D network. And the coverage can also be ensured by efficient utilization of the deployed BSs

Spectrum resource allocation and bandwidth enhancement
The challenge is to enhance the bandwidth to the extreme that the 6G requires to satisfy the user. The industry is talking about the terahertz frequency band to be used, which still faces problems such as weak coverage, immature terminal ecology and high construction costs Use of licensed and unlicensed spectrum while solving the network coverage issue by application of optimization mechanisms Perception and positioning Wireless spectrum in the 6G era can be used not only for communication but also for sensing and positioning functions. It is to provide communication and environmental awareness through base stations Location tracking services will enable a large number of emerging applications to get the precise location of the user's equipment

Network security
As the network is getting extended and more users are coming into coverage, and so are the emerging applications, network security is an important issue Post-Quantum cryptography (PQC) and Quantum Key Distribution (QKD) is being planned to be applied to the network to ensure network super security Green and low carbon This is an inevitable trend in the development of the ICT industry, along with all other industries following the same goal across the globe Enhancing the energy efficiency of the network by various optimization techniques as well as an efficient base station deployment one of the major development in 5G and beyond networks. The best way to fulfil the above performance goals is to reuse the entire frequency, that is, to reuse the frequency of one element. However, this compromises the efficiency of radio services and interruption management, resulting in heavy interference on a high-reuse item. Therefore, the RRM system should be developed to avoid distractions as much as possible over time, boosting spectral efficiency. The tight placement of small cells causes severe disruption and further complicates distributing of radio resources, so designing an RRM system for heterogeneous networks is extremely tough. Furthermore, due to an increase in small cells, the RRM system will need to expand and adapt for 5G and beyond networks. B. Interference Mitigation: For 5G and beyond networks, the size of the network is an ever-increasing one to meet the high number of user demands. So, there are more number of small cells to cover this. It is possible for a small cell to interrupt a neighboring eNB or other small cells in LTE and LTE-A networks. Different terminologies are used in literature to distinguish between different sorts of interference. In a two-stage femto-cell network, there are two sorts of disruptions: cross-tier and co-tier. Cross-tier interference occurs when femto-cells and macro-cells use a common channel. When the PRBs of femto-cells and macro-cells are the same, there is interference. For both the up-link and down-link, cross-tier disturbances are depicted in Figure 5. There has been a lack of attention paid to the potential for interference between the electrical network and the femto-electrical cell's system. Co-tier interference, on the other hand, is interference between femto-cells. Femto-cells spread uniformly within the macro-cell, and as a result, there is space between each of these cells, resulting in an evenly spaced cover. Using the same set of PRBs throughout a network of dispersed femto-cells can cause problems in both the up-link and the down-link. The interference scenario in scattered cells is depicted in Figure 6.
Intra-cellular disturbances, RN disruption, and cell disruption can be distinguished on supported networks (also known as multihop networks). 6,7 Direct links, backhaul links, and access links sharing the same set of PRBs can cause intracell interference. There is an inter-RN interruption when two nearby cells use the same PRBs. To describe

F I G U R E 5 Classification of RRM schemes for LTE/LTE-A networks.
the resulting interference as "intercellular disruption," the relay node (RN) in a macro-cell that shares PRBs with a neighboring macro-cell will use the term "nearby macro-cell MUE." DeNBs RN interference and RN interference are connected with interference between the two RNs. C. QoS Management: Quality of Service (QoS) is about managing users in the network to specify a guaranteed throughput level in a cost-effective way to improve user experience. QoS technologies aid in measuring and detecting any changes in the network conditions, such as congestion or availability of bandwidth. Thus to prioritize the user for 5G and beyond networks, the QoS is multiple times the present scenario. So, it is important to manage the QoS. 16 To manage it, there are plans to distribute femto-cells in public venues, such as airports, where the UE population is predicted to be large and the available radio services may not be able to match the QoS needs of each EU member. It is also considerably difficult to manage QoS in LTE or a variant of LTE. Each tiny cell in a network may only have a limited amount of frequency resources after the eNBs distribute them. 17 Consequently, to achieve the QoS standards of 5G and beyond networks, we need a more complicated RRM solution that considers interference barriers and restricted radio resources. D. Fairness: For 5G and beyond networks, fairness is an essential RRM issue. With regard to UE package planning, the conventional issue of fairness in RRM is that the proper amount of radio services for wireless access must be provided to each UE. The problem of bias in heterogeneous networks extends beyond programming to include the distribution of resources among the network's many small cells. This challenge is particularly significant as the accuracy of the resource separation between direct and backhaul access links should be considered in multi-hop 5G and beyond networks. 18 Integrity can be split into global and local justice issues in heterogeneous networks. In Reference 18, we first learn about the idea of LTE Fairness in resource distribution between direct access connections and backhaul-access, as well as package design among UE, are inter-twined in multi-hop networks. "Global justice", as used in this work, refers to the equitable allocation of resources throughout the world's many little cells. In other words, the distribution of radio resources is termed "global" if it fully meets the demand for specific cellular services. However, if the minimal level of access between UEs and radio executives is increased, radio resource allocation is considered very fair. Therefore, UE or low-level radio operators will obtain more radio services, whilst those with superior channel quality will receive fewer radio services, to put it another way. "High" local accuracy can be attributed to this fact. Performance measurements known as Jain's Fairness index 19 have been widely accepted in assessing fairness. E. Complexity: Complexity is another major concern as we are leading to 5G and beyond networks. The complex problems require complex solutions and algorithms which are affecting the latency and energy consumption in a practical scenario. Computing complexity and implementation complexity are the two subcategories of complexity. The RRM system's high signature value and the required information-sharing between BSs are two examples of the system's complexity. On the other hand, the RRM system's computer complexity relates to the amount of time it takes the RRM system to run particular algorithms in the BS. Conventional RRM methods may not be practicable in LTE / LTE-A networks in the near future due to excessive communication between BSs, which could cause inhibitory startup issues. Furthermore, in LTE/LTE-A systems, the maximum permissible computation time is hampered by the distribution of resources between UE being done across TTI. As a result, during the TTI period, the RRM algorithm's calculation time should be maintained. Overall, the RRM design is complex due to the requirement for users to take into account the implementation's complexity as well as the computing complexity.

USER ASSOCIATION IN HETEROGENEOUS NETWORKS
During wireless exposure to 5G, dense, heterogeneous networks may be the most common theme. However, the approach has certain drawbacks because of variations in the transfer capability of macro cells and smaller cells, which will lead to more user encounters with macro BS. 20 As a solution to this problem, 3GPP developed the notion of biased user interaction in Release 10 21 by illegally increasing user capacity gained from small BSs cells to assure that more users would be affiliated with small cells. For bias user organization, the advantages of off-loading release from macro to tiny cells were demonstrated in References 22 and 11 through enhanced accessibility. Users who are forced to associate with smaller cells because of added bias get substantial disruption from the surrounding macro-cell. In this case, a significant disruption could undo the gains made by packing traffic into fewer cells. There is a high correlation with the number of biases adopted, which should be carefully constructed in order to maximize network use. 23,24 In Reference 13, Q-learning was used to determine the bias of each user, where each user learnt independently from the prior experience how much bias lowered the number of users in the termination. Inter-cell Interference Co-ordination (ICIC) and enhanced inter-cell interference (eICIC) are two resource-based disruption mitigation strategies developed by 3GPP to address the aforesaid problem of biased user integration. Both the bias rate and the segmentation of resources in eICIC-enabled heterogeneous networks were established by the authors. 4,25-27

User association for coverage or outage probability optimization
For 5G and beyond, networks are to be tested for coverage or outage scenarios to see how well a user performs. Using stochastic geometry, outage or coverage capabilities are employed as critical performance metrics in user organization analysis. Researchers in References 28 and 29 used stochastic geometry to describe and examine the max-RSS user association's performance in K-tier down-link heterogeneous networks. The K-tier heterogeneous network's cell load was shown to be heterogeneous, and the potential for low interference interfaces for diverse networks was established in Reference 12. Some network features may not be able to contribute to the amount of integrated disruption because of huge loading disparities across current network pieces, according to the authors in Reference 12. Because of this, the SINR model 12 was established in Reference 25 to monitor the activity factor of different BSs. Femto-cells and pico-cells, which are less densely populated, can improve network coverage. There may, however, be congested network features as a result of the random transmission of small cells paired with considerable variances in transmission power relative to large base stations. A similar strategy to References 30-32 was taken by Reference 13 authors, who started by identifying coverage chances for each category under different spectrum distribution rules and femto-cell access regulations before discussing the volume-expanding problem. These are some methods used recently in the literature review to overcome some of the QoS hurdles found in terms of both available coverage and the lower levels of each category. The findings gave a clearer picture of the full spectrum.
In the context of multi-tier down-link heterogeneous networks, stochastic geometry was used to examine the impact of bias user interaction. Total bias leads to a greater signal-to-interference (SIR) ratio. The deepest level of integration was uncovered through the use of statistical analysis. In References 5,24,33, smooth user interactions and spectrum separation between macro-cells and smaller cells were considered. According to Reference 20, the two-phase topology serves a specific purpose and provides information on the best spectrum classification based on a statistical study. It is more likely that 33 inputs that assure specialized functionality will be available in a normal multi-stage network when the network has low connectivity. User bias and spectrum partitioning ratios were calculated to maximize equitable network utilization based on the potential for both heterogeneous network down-link and up-link, as opposed to the previously specified functions of down-link heterogeneous networks. A trade-off between up-link and down-link functionality on heterogeneous networks, where users are compelled to associate with both an up-link and down-link BS, was revealed by the results to exist between the total user organization's up-link and down-link biases.

User association for spectrum efficiency optimization
It is commonly agreed that spectrum efficiency is a useful statistic for network performance in 5G and beyond networks. In order to enhance the total number of users, a variable user relationship was proposed in Reference 34. An upper limit on total down-link rate was discovered using convex optimization, and then a heuristic user association law was proposed that was simpler and closer to higher performance than the convex optimization approach. Compared to previous RSS and biased user data, their simulation findings showed the proposed heuristic user interaction is superior. It is widely accepted that increasing the overall quantity of data available to all users may lead to erroneous data allocation. However, the authors in Reference 34 overlook the up-link factor, which might hamper the QoS considerably. Smaller cells tend to clump together because they bear a greater burden than macro-cells. Since only lucky macro-cell users are able to access high data rates, the rest of the users are left feeling hungry. Users' data rate was defined as the sum of their logs, hence a complicated low-level method for organizing users was devised to maximize this rate. 35 Providing greater resources to low-level users is preferable since the logarithm is a concave function with a low return, thereby promoting both load balancing and user bias in the system. In Reference 36, significant unresolved difficulty in compound growth was turned into a convex developmental state by reducing the primal deterministic user association to the fractional association. Problem convexity helped to build the distributed user algorithm using dual decomposition, gradient downtime, and a guarantee of not exceeding specific high deviations from correctness. The size of the step has a significant impact on the assembly speed of the gradient drop approach. Using the same problem as in References 37 and 38 offered an integrated descending strategy to provide a more reliable performance guarantee and faster integration. To meet the proven performance duty, the integrated problem of minimizing service delays required by users was developed in Reference 28 and solved with the use of measurement techniques. Nevertheless, the authors in Reference 28 state that the major drawback is that the system is not aware of the load variation and again does not consider the up-link scenario. Moreover, the system model is also static rather than a dynamic one.
It is also common to employ game theory in the context of spectrum utilization. User association in heterogeneous networks can be thought of as an open-ended challenge in References 15 and 39, where each network functions as a competitor by providing users. Negotiation was created to maximize data utilization, while ensuring specific levels of users, as well as providing fairness for all users and balancing cell traffic load at different phases. In order to preserve the quality of service, multi-resource traffic 34 created an opportunistic user algorithm to categorize individual traffic as a primary service and machine-to-machine mobility as a secondary service. Adequate distribution of services for the second service by the planned exploitative organization will not compromise the core QoS service. This is a multi-to-one simulation game in which users and base stations evaluate one another based on well-defined user resources in down-link user association on heterogeneous networks. 5,12,33 According to precisely linked essential aspects in Reference 40, users and BS rank based on specialized application functions that calculate both data rate and accuracy for cell edge users. In contrast to Reference 10, a service-based user organization was built by focusing on differentiating users' priorities, delivery time, supply failures, and the diverse Quality of Service (QoS) need of users. The authors of Reference 17 employed a problem-solving approach similar to the one in Reference 41 to illustrate the quality of user information in terms of mean concepts that accurately reflect specific properties of the investigated 5G and beyond network applications. It is possible to persuade a small cell to take traffic from a macro-cell in order to enhance its coverage. However, all of the authors mentioned in this section above avoided the interference issue in general when considering their separate objectives.
In heterogeneous networks, the requirement to manage interference is exacerbated by overcrowded tiny cells. Research initiatives have made considerable gains in the integration of user interfaces and other areas of the efficient distribution of radio resources. To maximize the amount of data, the evolution of user interaction and channel allocation decisions between macro-cells and tiny cells was studied in Reference 22. As stated in Reference 33, user integration, 9 connection coordination and channel allocation between macro-cells and smaller cell groups were recommended to maximize data utilization. The reduction of heterogeneous networks 7,20,27 and the increase of heterogeneous networks 42,43 have been studied in relation to user power control interaction. Users' organization and power transfer are both under constant scrutiny in the proposed algorithms 37,44,45 until integration is achieved. As a non-participatory game, the authors of Reference 46 have raised the issue of raising the amount for both users and BS to play. In order to increase thermal conductivity, it was proposed in Reference 47 to construct a network of small interactive cells that combined user integration with spectrum distribution and interference interactions with surrounding cells. Because of the NP-hardness of the previously described shared development for user integration and channel or power control, finding the proper solution is no easy job. For example, the solution can be approached by sequentially examining the user's affiliation and power level until an agreement is established. 37,44,45 Fixed channel allocation or transmission coordination can help increase user engagement before optimizing channel transmission coordination either way. 4,29,31,33 With this in mind and the importance of user association in mind, we can say that careful cell selection optimization is critical for the holistic optimization of heterogeneous networks.

6.3
User association for energy efficiency optimization 5G and beyond networks will inevitably consume more power as the volume of data traffic increases and the network infrastructure expands. To qualify as environmentally friendly, this will immediately increase glasshouse gas emissions. As a result, businesses and experts are simultaneously working to increase network capacity and energy efficiency. Network capacity can be increased by increasing the amount of data successfully transmitted while reducing total power usage. The reduction in overall power consumption while meeting corresponding traffic requirements or the increase in the ratio between the total data usage of all users and the total network power consumption can both be viewed as boosting network efficiency in terms of problem-solving. When a user is connected to a macro-cell, the power consumption of the access network is typically higher than when connected to a smaller cell. A user association's decision on how to allocate network capacity, therefore, has a significant impact on its efficiency. Energy savings on heterogeneous networks are made possible by numerous significant contributions published in user organizations. 2,3,40 To increase system efficiency in heterogeneous networks with high user power transfers and low-level restrictions, a user organization algorithm was proposed in Reference 16. User engagement in heterogeneous networks was improved by raising the ratio of data for all users to total power usage in Reference 45. In contrast to Reference 45, where the focus was on problem-solving, 10 writers studied how to reduce total energy use while still meeting the needs of traffic users through user engagement. In order to improve the peak signal-to-noise ratio, the authors of Reference 7 advocated that video content recognition of user interaction be used to decrease heterogeneous networks measuring and power consumption of the system. Thus, both indirect classification and dual decay techniques are employed to tackle the problem. In order to maximize down-link output, algorithm-based decomposition was devised in References 26 and 4. In order to accomplish this, users were assigned to a specific BS, which reduced the overall amount of transmission power used.
An analysis of mobile communication systems shows that radio access nodes account for 57% of the total energy consumed by wireless networks. 1 In addition, signal processing and air conditioning circuits consume about 60% of the energy spread in each BS. 34 Thus, blocking base stations that do not support active users is an efficient method of reducing network power usage. 14, 35 An attempt was made in Reference 28 to boost energy efficiency or minimize total use while maintaining a fixed rate and accuracy ratio by preparing BS sleep mode for long-term sleep, organizing users, and allocating sub-carriers jointly. Using computer simulations, these two strategies (raising energy efficiency and reducing total energy use) were tested. As described in Reference 24, an algorithm was developed to reduce energy consumption by optimizing the functionality of both the user interface and BS sleep mode to take into account spatial and temporal variations in traffic demands as well as internal hardware components such as BS. Using a stochastic geometry-based model, 17 the potential for coverage and power efficiency of K-tier heterogeneous wireless networks was discovered under various sleep modes. The authors of this work have found a way to both minimize energy usage and expand the amount of energy available to macro-cells. Table 1 represents a brief summary of algorithms, parameters and their pros and cons. There are some emerging issues with these heterogeneous networks. The inherent heterogeneity of heterogeneous networks presents itself in terms of up-link-down-link asymmetry, the backhaul bottleneck, varied footprints, and so forth, aside from the transmit power imbalance between small cells and macro-cells. The design of the user association faces significant difficulties because of this. However, the majority of prior research has merely mentioned these difficulties in passing. Therefore, we will focus on three key points in the following paragraphs:

Down-link-up-link asymmetry
Most research on user interaction on various networks has investigated the problem from the up-link or down-link view. However, various networks often introduce asymmetry between down-link and up-link depending on channel quality, traffic volume, coverage, and computer hardware limitations. Among them, down-link and up-link coverage asymmetry is very difficult for various networks. In the down-link, macro-cells provide much greater coverage than smaller cells because of the large variances in power between different types of primary channels in various networks. In contrast, user resources can transfer the same amount of power to the up-link regardless of the type of BS they are connected to. Although some promising results have been reported in References 16,41,48 regarding down-link cuts and up-link user associations, these cuts obviously require excellent synchronization and high-speed data connection as well as minimal latency between basic channels in order to be effective. Users have been required to associate with the same BS in both up-link and down-link directions since the invention of the mobile phone, 39 as this integration makes it easier to construct and employ sensible transport, and physical channels. As a result, 5G networks are likely to utilize both down-link and up-link technologies. User-friendly links may not perform correctly because of network coverage asymmetry in up-link and down-link. There are many ways that a big RSS-based user association law could relate a remote macro-cell to a user's location. For tiny cellular users, this results in heavy up-link interference, which reduces spectrum efficiency and power, and shortens battery charging time, because the user must transmit excessive power to ensure the intended signal intensity gets up-linked. In order to segregate down-link and up-link submissions across multiple networks, it is noticed that asymmetry remains, whether Time Division Duplexing (TDD) or Frequency Division Duplexing (FDD) is employed. As a result, in heterogeneous networks, using a complex joint up-link and improving user engagement is essential. The authors of Reference 49 presented a user association method that raised the number of allowed users while reducing the limiting power consumption of up-links. In Reference 25, the algorithm's performance is highly dependent on the weight, which was determined heuristically. The objective function in Reference 25 was further enhanced to increase network utilization, which was based on a trade-off between data rate and power consumption. However, the authors in Reference 25 consider base stations to have full queues and therefore, biasing deteriorates the outage rate of the overall network by lowering the signal to interference ratio (SINR). Improved user experience was developed as a discussion problem in Reference 50 to boost up-link and down-link efficiency resources. Under specified Quality of Service (QoS) restrictions, a user association and a beam-forming algorithm were developed in Reference 42. Table 3 presents the summary of algorithms, parameters and their advantages and disadvantages in heterogeneous networks.

Backhaul bottleneck
We can expect a paradigm shift in research due to the emergence of heterogeneous networks. The importance of the backhaul in the context of the 4G LTE network was not completely acknowledged among these difficulties. Moreover, it becomes more obvious in 5G and beyond technologies. 45 Even if a complete backhaul is used, most studies focus on front-end performance without considering specific backhaul implementation specifics or any potential bottlenecks in the system's network controller. This is often true for well-structured macro-cells from the distant past. However, BS may slow down traffic in diverse networks due to overcrowding. Due to their restricted data rate, current backhaul options for small cells are not ideal. When considering dense heterogeneous networks, rigorous backhaul analysis is necessary to realize their full potential. For heterogeneous networks, the backhaul volume limit is quite critical. As a result, backhaul-aware interaction techniques that consider the backlash limit are necessary for heterogeneous networks. In order to maximize the efficiency of a wide range of networks, the dispersed organization algorithm uses a flexible integrated setting. 4 Under reversal limitations, several user-based metrics were examined. A new algorithm was introduced in Reference 5 to enhance the overall weight of all users in conjunction with carrier aggregation (CA) and to force backhaul capture in smaller BS cells. A heuristic user association technique was developed by the authors of Reference 46 to boost network capacity when both backhaul volume and cell load were considered. Using the power of game theory in backhaul-constrained heterogeneous networks, a cache-aware user association was constructed, which has been categorized as one to many simulation games. Based on the volume and recognition of a device that estimates the BS data storage capacity and user movement patterns, it has been found that BS users are linked to each other. For a far lower price than assured QoS connections, the authors of Reference 48 proposed an intriguing approach in which third parties provided backhaul connections for BSs and then leased their networks' excess capacity to mobile service providers. The authors in Reference 48 give a standard algorithm for optimizing user engagement, which allows mobile carriers to strongly select which customers should be outsourced to third-party femto-cell based on common traffic requirements, interruption rates and channel conditions as well as third-party access pricing. When optimizing the backhaul network, the authors 15 took into account the lowest number of users while allocating resources across all wireless channels and controlling the flow. In terms of network power usage and backhaul energy, researchers 21 read the interaction stories of heterogeneous networks' energy efficient users by considering both network power consumption and backhaul's energy.

6.4.3
Mobility support The increase in cell density found on heterogeneous networks for 5G and beyond networks poses ongoing challenges for mobility support. Small cells have reduced transmission capacity, leading to a reduction in footprints. A static method that does not take into consideration the user's mobility may result in a common dilemma between cell association on heterogeneous networks compared to a conventional equivalent network. On the other hand, it is well known to increase the complexity of complex procedures, resulting in higher expenses and unnecessary delays in the delivery and withdrawal of calls. 47 The load balance and handover process of heterogeneous networks is often not as good as systems with pure macro-cell (homogeneous) only systems, as noted in the 3GPP technical study. 16 In order to optimize long-term system performance and avoid over-sharing on heterogeneous networks, it is vital to account for user transactions when making user interaction decisions on heterogeneous networks. The authors of Reference 11 initially discovered the opportunity to cover down-link by considering user speed under a biased user integration law by taking advantage of stochastic geometry. After that, there was a total bias in favor of expanding coverage, with both a positive bias and the likelihood of a merger being linked to the user's speed. The findings in Reference 11 showed that the bias factor-based speed factor was able to increase both network coverage and overall network performance. According to Reference 12, a modified Markov and Poisson process 51 replicated the user's movement and was used to improve system performance in terms of traffic delay and blocking potential. While the authors in Reference 13 presented that old algorithms were only transmitting one package at a time to either the BS or to a mobile TA B L E 3 Summary of algorithms, parameters and their pros and cons in heterogeneous networks. user with a negative delay on heterogeneous networks with two subcategories. As a result of the erratic nature of interactions between BS and users, the need for a multi-point transfer strategy emerged, in which many BSs keep useless copies of data and are grouped such that mobile users may receive data reliably. For the authors, 15,46 a novel framework was devised, using a new technique to find the best user association in the femto-cell network, which not only assessed previous cellular behavior and anticipated future regions, but also reduced the frequency of delivery. Due to network heterogeneity, 3GPP Release 12 has temporarily suspended dual communication. It is possible to simultaneously connect to both macro and small cells via dual connections, which considerably improves mobility endurance and increases user access to the network. As long as dual communication is enabled, the user will be able to receive signed messages from the primary BS. This user does not need to begin rendering processes until they move to the cover of another huge BS, thereby ensuring more efficient delivery. User interfaces determine in advance the benefits of dual-connection functionality, which determines which BS the user should be linked with. Simulations have been used to examine the impact of various user-related circumstances on the ability to achieve dual connectivity in Reference 18. In Reference 19, the total number of users was varied to analyze the cell association, and a sub-optimal algorithm for creating sub-optimal associations was presented to resolve the issue. However, a dual connection has the ability to improve user mobility, but dual communication also offers a number of technical issues in terms of buffer status reports and reporting, transmission power management and so on. Because of this, extra work must be done before dual communication can be completely utilized to promote mobility across diverse networks. A summary of the qualitative comparison of the most common user association algorithms for heterogeneous networks is presented in Table 4.

TECHNICAL APPROACHES FOR LOAD BALANCING OR FAIRNESS
When measuring the burden on various servers and equipment, load balancing has long been considered a means to boost values such as resource utilization, accuracy and wait or process delays, or exits. Even if the distribution of users is uniform, "natural" measures like SINR or RSSI can lead to severe load imbalances in emerging 5G and beyond networks because of the varying transmission power and BS capabilities and the ever-increasing number of users. Due to increased network traffic and network outages, a broader system needs to be developed in which decisions such as user configuration and mobile connectivity are integrated. To discover the most efficient user server structure is a complicated challenge, and the complexity grows exponentially with the network rate, which is a dead-end approach. In the next few short sections, we will go through some of the most useful technical approaches of load balancing.

Relaxed optimization
This NP method is difficult to use, even for a tiny mobile network, due to the more constant relationship between user organization and planning that emerges from a typical user rate less than a service and power limit. As traffic increases, so does the difficulty of solving the integrated lines problem, which has been extensively studied in line theory with only a limited amount of improvement over time. An alternative method for making the problem more convex is to allow users to associate with many base stations in a full-fledged model that regularly transmits. 3 Depending on the application and power limit, the fundamental form is to increase the weight level of the load, where the binary organization indicator is relaxed to a real value between 0 and 1. An algorithm that is both simple and near-optimal can be created using the most common programming tools, such as double decay, which is followed.

Game theory
Using game theory as a field, researchers can analyze collaborative decision-making processes and study the most pressing issues of spatial expansion, 34 analyses, for example, a user-focused method that does not necessitate any top signature or connection of several access networks. Users in different service areas compete for bandwidth from many wireless networks in the study of network selection variables. 25 A valuable tool for operating systems and flexible networks, game theory is not guaranteed to result in a combination of algorithms. Even after combining the methods to get a better result, this may lead to incorrect use due to the high cost. As a result, no closed assertion can be made about the link between performance indicators and network parameters in game theory. Game theory can provide some insight into how unmanned vehicles and base stations should be coupled in terms of network load fluctuation.

Markov decision process
Learning about the evolution of stochastic systems of indefinite time is possible using Markov decision-making processes (MDPs). Mobile uploading to WiFi has been studied using MDPs in the setting of heterogeneous networks. Another interesting use for heterogeneous networks is the challenge of organizational structure; for example, Reference 35 presents a hybrid scheme in which users are aided in their decisions by streaming data. But as network size increases, direct resolution of MDPs becomes increasingly challenging. Using MDPs in a continuous state space also has its drawbacks. There is also a challenge in defining appropriate circumstances and a coherent change model in complicated, unstructured situations. It is difficult, in general, to establish a proper nation model and solve it precisely on a big diversified network that includes various types of BSs and WiFi, but MDP provides a feasible technique to think for itself, combining the advantages of a medium and a distributed-design network into a single network.

Cell range expansion
One of the most prevalent sub-optimal techniques for offloading users to reduce power BSs in 3GPP standardization efforts is the use of biased received power-based user association control. 44 This strategy uses an association bias to unload users to smaller cells. If a person can choose from among k possible tiers, the index of the tier they choose is where B i is the bias for tier i and P rx,i is the received power from tier i. Scaled-down cells are placed in Tier 1 of the cell hierarchy (0 dB). In other words, a small cell bias of 10 dB means that until the UE device's received power drops below 10 dB below the macro-cell BS, it will remain connected to the small cell. Biasing has been dubbed "Cell Range Expansion" (CRE) because of its ability to increase the range and area covered by small cells. It's common to reflect about the appropriate distance between CRE and the theoretical solutions that were previously described. The simple bias in each phase virtually reaches optimum load performance when fair values are chosen. However, in general, it is difficult to identify which strategies to use effectively are biased. 42

Stochastic geometry
For the current network setup, the utility function U is maximized by the use of the prior tools and schemes.
where Ω is the set of solution space. However, if the network configuration is based on an underlying distribution, a different problem can be formulated as in Equation (3), where the optimization is over the average utility: This is a stochastic optimization problem, and hence falls under this category. Using Equation (3)'s solution for Equation (2) would be a mistake. As a result, the gap between a well-designed but stable CRE and the world's best solution has been proven to be substantially smaller than if the organizations of each network were reorganized.
As a branch of possible opportunities, stochastic geometry provides BS and user locations in the network via a point process. The Poisson Point Process (PPP) can yield expressions from important metrics like SINR and scale 50 when modelling user locations with BS. It is also valuable for providing information on the influence of important system level factors like transmission power, density and bandwidth of different categories when constructing load balancing algorithms. Stochastic geometry 20 was used to study cell expansion and found to have a significant effect on the network parameter in a concise manner by measuring overall potential network configurations.
As a result, the organization's precise location and distribution are unaffected by modelling BS as random sites in several networks. According to Reference 20, an organizational spatial measuring method was provided, which was then utilized to distribute the load (based on the same distribution of users); as a result, the bias parameter distribution for each category may be determined. 20,21 It is possible to utilize the rate distribution to calculate the correct bias by adjusting the distribution ratio to increase as a function of the bias value.

7.6
Load balancing accommodating other emerging issues Load Balancing for 5G and beyond networks is far from fully understood. New flexibility and benefits for system designers are obvious, but it also requires a lot of generally used metrics and ideas that have been created through time to locate similar mobile networks. As a result, new approaches to study and evaluation are being considered. 7.6.1 Backhaul bottleneck Wired backhaul connections often limit the power of small cells, such as femto-cells or WiFi APs. Small cells loaded across the border, which may be retrieval-dependent, might limit the quantity of unnecessary data extracted from macro-cells in light of this backhaul strain. 21 The simplest approach would be to add the bias value to the backhaul limit.

Mobility
Supporting seamless supply between different cell types in different networks is important. When a user with moderate or high mobility encounters a small area of cell structure, the user must be removed from their original cell and return when the small cell is no longer nearby. On the other hand, handovers are well-known for their lengthy and expensive procedures. Short connections to small cells may necessitate tolerance of BS integration that is less than optimal from a system-level perspective rather than commencing supply and departure from the cell. Open versus small closed access cells are a similar issue.

UE capability
In addition to its obvious advantages, bias in small cells lowers SINR. LTE-A systems can achieve zero link throughput under Adaptive Modulation and Coding (AMC) if the SINR is less than −6.5 dB. Since the UE device may theoretically have a better rate than the cell with the same SINR (−15 dB), this is not possible since the cell's low bandwidth and coding prevent it from doing so and edit BS can be sent. This can help with management and disruption cancellation, but it also adds a new hurdle to precisely estimate the cost of load balancing.

Asymmetric down-link and up-link
In the down-link, macro-cells cover a bigger area than small cells due to the substantial variances in strength between BS types in different networks. On the other hand, UE devices can transmit the same amount of power regardless of the kind of BS. down-link traffic, on the other hand, tends to be significantly more intense than up-link traffic. With these inconsistencies, a full down-link link does not need to be important for up-link transmission. As a result, the current down-link load balancing algorithm must also be extended to include the relevant up-link cases. Ideally, researchers should conduct a collaborative study to quantify both down-link and up-link traffic.

CONCLUSION
The LTE or LTE-A architectural framework has gradually changed into heterogeneous networks over time. High-speed wireless communications will become a reality as a result of this development. In order to increase capacity, 3GPP uses LTE-heterogeneous network deployment as a means of increasing capacity. However, the most challenging difficulties in heterogeneous networks where small cells are layered on top of macro-cells are cell selection, load distribution, and interference mitigation. Resources, complexity, and the Quality of Service (QoS) are just a few of the RRM issues preventing these networks from progressing further. These problems are being addressed through a variety of methods. Furthermore, the dynamic nature of user distribution and service demand means that the ultimate goal of improved cell-edge and overall spectral efficiency and QoS cannot be accomplished. This study provides a comprehensive evaluation of the relevant literature. This literature review includes several adaptive approaches that can boost each of these characteristics independently to an improved degree. Despite this, heterogeneous networks remain unreliable and insufficient for 5G, 6G, and future ultra-dense networks. The lack of a unified optimization of QoS, capacity, and spectral efficiency is to blame for the lack of progress. The use of multi-objective frameworks and hence the optimization of the level of satisfaction for all areas is one possible method. Research on heterogeneous networks is increasingly using this method. Examples include game theory, genetic algorithms, artificial intelligence-based adaptive methods, and others. Designing in this manner is intriguing because it has the potential to be quite effective in meeting numerous objectives at once. To summarize, the research topic of heterogeneous networks still faces many challenges, some of which have been discussed in this paper. By overcoming these challenges, LTE or LTE-A networks will be able to evolve into heterogeneous networks and enable 5G, 6G, and beyond. A multi-objective-based study in heterogeneous networks is recommended for its advancement.

ACKNOWLEDGMENT
This research work has been supported by the Australian Government Research Training Program (RTP) scholarship along with CQUniversity, Australia. Open access publishing facilitated by Central Queensland University, as part of the Wiley -Central Queensland University agreement via the Council of Australian University Librarians.

DATA AVAILABILITY STATEMENT
The data supporting this study's findings are publicly available via the Google Scholar engine at https://scholar.google. com/.