Latency optimization approaches for healthcare Internet of Things and fog computing: A comprehensive review

1 Introduction

The Healthcare Internet of Things (H-IoT) is rapidly transforming traditional healthcare systems by enabling real-time monitoring and diagnosis [1]. H-IoT connects smart devices, wearable sensors, and cloud services to create healthcare ecosystems that enhance patient care and administrative policies. However, if not effectively managed, this technology can introduce latency in data processing and transmission, which is unacceptable for time-sensitive healthcare applications such as emergency response, telemedicine, and remote surgeries [2]. Reducing latency allows healthcare experts more time to provide accurate responses, directly impacting patient outcomes. According to the International Data Corporation (IDC), the number of connected devices is estimated to reach between 41.6 billion and 1 trillion by 2025, generating an unprecedented volume of data measured in zettabytes. This trend is driven by the proliferation of IoT devices, the growing use of social networks, mobile applications, a rapidly increasing global population, and a technology-driven lifestyle [3].

In healthcare, IoT devices produce vast amounts of multimedia data that require effective management. Cloud servers worldwide handle these data, including analysis, storage, and preprocessing. The sheer volume of data has significant implications for healthcare, making it crucial to extract meaningful “signals” from the “noise.” Cloud computing plays a vital role in connecting IoT devices to the healthcare industry. Data generated by IoT devices is typically processed, filtered, preprocessed, and aggregated using cloud services. However, cloud computing has limitations, particularly as data transmission rates increase. Issues such as high response times, service latency, and data loss become more pronounced, degrading the quality of service (QoS) for end-users. These challenges are especially critical for IoT systems requiring real-time data processing, such as healthcare and city management systems. Healthcare systems rely on real-time data processing to address urgent situations promptly. The current service model, which requires IoT devices to connect to routers and gateways before reaching the cloud, introduces significant computational latency. The greater the distance between IoT devices and the cloud, or the more routers deployed in the network, the longer the data transmission path and the higher the bandwidth consumption. These challenges highlight the need for solutions that minimize latency and reduce network bandwidth usage to meet the real-time demands of healthcare IoT systems [4,5,6].

The integration of IoT with fog computing (FC) presents immense potential. To fully realize these possibilities, sufficient networking infrastructure must be provided to maintain low latency for IoT applications. FC plays a critical role in the execution and processing of data from IoT devices. It offers a robust solution to address these challenges by enabling applications to run and data to be processed closer to the IoT devices that generate the data. This approach is particularly effective in overcoming the limitations of centralized cloud computing and enhancing the efficiency of IoT systems. FC emerges as a promising solution to latency issues in Healthcare IoT systems. By decentralizing and distributing computing resources closer to edge devices, FC reduces reliance on centralized cloud services, enabling faster data processing and decision-making. It also helps alleviate network traffic and improves the overall availability of healthcare services. However, integrating FC into H-IoT architectures introduces additional complexities, particularly in resource management, task allocation, and ensuring data privacy while minimizing latency. Addressing these challenges is essential to fully leverage the benefits of FC in H-IoT systems [7,8,9].

This review systematically identifies and discusses the state-of-the-art approaches regarding the latency optimization of H-IoT systems using FC. It analyzes architectural improvements, resource management strategies, and communications processes aimed at improving system performance. Besides, it enumerates the current challenges and open issues, which include energy consumption, scalability, and the trade-off between latency and security. Based on the synthesis of recent advancements, this work seeks to review and understand latency management in the context of fog supported H-IoT systems to inform future improvement of similar healthcare technologies. The main contributions of this study are:

• This work provides a comprehensive overview of the sources and impacts of latency in H-IoT applications. It highlights the critical importance of low-latency solutions for expanding healthcare use cases, including effective telemedicine, real-time monitoring, and emergency response systems.

• The study explores how FC addresses latency challenges in H-IoT systems by distributing computation and resource-intensive tasks closer to data collection points. Quantitative analyses support qualitative insights, emphasizing the architectural and operational benefits uniquely suited for healthcare environments.

• The work assesses current strategies for task scheduling, resource allocation, and optimized communication protocols in fog-based H-IoT systems, focusing on their effectiveness in reducing latency.

• The review identifies limitations in existing FC approaches for latency minimization, including energy consumption, system capacity, and the latency-security trade-off, providing a foundation for future research.

• By consolidating knowledge from current research, this systematic literature review offers actionable recommendations for designing and implementing latency-sensitive H-IoT systems, aiding healthcare administrators and system designers in their efforts.

The rest of this study is organized as follows: Section 2 provides the requirement of Healthcare IoT Systems. The basic of FC is presented in Section 3. Section 4 provides the related works on latency reduction. Existing algorithms and techniques for latency reduction in cloud and IoT using FC are discussed in Section 5. Section 6 presents minimizing packet loss and data traffic. Techniques to achieve QoS requirement for IoT is discussed in Section 7. Section 8 provides FC performance tools. The gap and challenges are discussed in Section 9. Finally, the conclusion and future works are presented in Section 10.

2 Requirement of healthcare IoT systems

By 2030, 750.5 zettabytes of data will be created by billions of IoT devices. To efficiently compute and transfer data, reducing latency is essential for improving cloud response and service time [2]. However, end-users may still experience increased latency due to the physical distance and network congestion between devices and the cloud. In this context, latency is routinely measured and analyzed. It refers to the time taken for data to travel from the source (point A) to the destination (point B). Total cloud latency typically comprises three components: processing delay, transmission delay, and queueing delay [3]. The total latency in the cloud is the sum of delay throughout the transmission channel, which includes the time a packet spends on the wire or fiber, as well as processing time at each intermediary node. As a result, healthcare IoT requires minimal network, service, and computing latency for real-time applications. The proliferation of IoT devices has resulted in the creation of massive amounts of data. As a consequence, there are a large number of requests for related services from users via the network. For real-time inquiries, most IoT applications demand low latency and quick responses. The advent of IoT in recent years has created tremendous worry among services that require real-time data [4]. The probability of making a mistake is related to the magnitude of the data transmission. As a result, end-users may suffer poor QoS. End-users faced unacceptably high transmission latency and poor services as a result of the massive volume of data created by IoT devices [5].

The difficulty of synchronization between a client's request and the server's answer is exacerbated by high service latency. Furthermore, excessive latency in clouds is caused by network congestion between IoTs and clouds. Communication between IoT devices, the cloud, and end-users are always multi-hop. To be able to forward data packets to end-points, gateway nodes are required. They are determined by the distance between the remote server and the number of routers used [6]. For information bundle sending to end-focuses, the more extended the distance, the more gateway hubs are required. Besides, the method is deferred in the event that the parcel does not stream straight by means of the switch to arrive at the end-point; the bundle might need to go through a few switches. As a rule, switch parcels were deferred by a couple of milliseconds, bringing about a couple of milliseconds for each bundle full circle time. These switch gateway hubs create latency because of the basic information calculation delay. The proficiency of the network and the nature of the steering gear influence reaction times to end-client requests. Information duplication to many cloud server farms in different trans-mainland areas would be deferred too. These deferrals because of high network latency are an immense worry in the IoT-cloud framework, since the network state is continually changing with time [7]. Network latency [7] is the time it takes for a message or communication to move between various locations.

For software applications that must function on a global scale or high-performance software systems that require a very quick response time, network latency accounts for a considerable amount of the necessary reaction time. The unpredictable nature of the network leads latencies to be unpredictable. When data are transported from IoT to cloud servers [8], it undergoes a new life cycle. The quantity of data stored on servers and devices reduces their energy consumption. As a result of the high latency, data traffic, energy, and network consumption, the healthcare IoT-cloud system becomes incapable of providing real-time services to patients and clinicians. Therefore, there is an opportunity and urgency to engage in this area.

3 FC

Whether centralized or decentralized computing models should be used has been argued in recent decades based on two primary sets of difficulties. To begin, consider efficiency vs effectiveness while seeking answers using a rationalistic technique. Second, there are the politics of organizations and resources. At the turn of the century, the centralized cloud computing model was established [9]. To avoid excessive latency, location awareness, and other issues that may arise in a centralized method, it is necessary to move computing from centralized to a unique decentralized technique. FC was coined as a result of the shift from centralized to decentralized computing [10].

FC presented itself as an infrastructure design by Cisco Systems in 2012 [11]. Fog operates across three different levels within network systems according to their definition. The three main characteristics of data collection under FC include sensors and vehicles and ships and roads while also processing the data faster than one second for decision-based measures [12]. FC allows cloud services to approach end-users through distributed compute resources and storage facilities while providing enhanced security together with mobility features and low-latency performance and increased privacy and network bandwidth. Applications that require real-time processing receive significant advantages from using FC platforms [13]. Fog layer nodes possess sufficient processing capabilities which enhance both energy efficiency of the system's sensor nodes along with data transmission performance [1]. A system-level architecture named FC consists of multiple nodes, which form its primary main components that deliver cloud-based features. Different connection systems including both wireless and wired enable fog nodes to form a network. One key feature of FC allows networks to establish connections with the cloud. The OpenFog Consortium Architecture Working Group (2017) states that FC puts mission-critical data-intensive applications ahead in their network infrastructure by moving functions such as communication, storage, computation, and control to data creation locations at network edges. Three main characteristics describing FC are as follows:

• Low latency and location-awareness: Quality services significantly benefit from FC because the network edge lacks sufficient support. FC brings low latency behavior and geographical diversity through its network-edge position by bringing cloud control and processing capabilities closer to local devices [14]. FC fulfills the timing requirements that emergency and healthcare services need according to research by Ala’anzy et al. [15] and Baskar et al. [16].

• Geographical distribution: FC creates a dispersed network topology through its deployment structure of distributed fog nodes, which differs from centralized cloud distribution. The spread nature of fog layer deployments enables the system to distribute data analytics and processing throughout the network thus enabling it to deliver fundamental services together with locally managed distributed systems that provide excellent quality services [17].

• Large-scale sensor network: FC supports large-scale sensor networks through professional management of end devices by adding or subtracting nodes to optimize latency and response time [18].

• Mobility: FC uses an architecture design which exhibits built-in support for mobile applications through all-inclusive management of IoT and mobile sensor functions. The solution requires devices to connect directly with fog nodes so that cloud servers become less essential for operation. The primary strength of FC consists of its mobility features since it allows devices to move across networks without demanding reconfiguration procedures. The technology benefits dynamic environments like vehicular networks, smart cities and industrial automation because moving devices require no reconfiguration. The network edge implementation of data processing through FC yields both low-latency performance and high-speed responses that are essential for applications needing real-time decisions. The fast processing of traffic data by vehicles' mobile sensors, which communicate with local fog nodes, allows both better traffic management and decreases vehicle congestion. FC suits applications, which feature mobile devices because its high mobility level provides an optimal framework [19].

• Heterogeneity: FC platforms naturally combine diverse elements because they provide various storage and network and computational services. These services undergo virtualization and deployment through a layered system between end devices and the cloud to support diverse environments. Absolute diversity marks fog computing by letting its structure unite different kinds of devices and technology protocols together. The variety of fog nodes encompasses routers as well as gateways and access points and specialized edge devices, which serve applications in different environments ranging from urban to remote industrial locations. The adaptable nature of FC lets it support numerous applications which range from domestic settings to extensive industrial IoT (IIoT) platforms [20].

• Real-time interaction: Due to its nature, FC stands out as an excellent choice for time-sensitive applications that need instant processing abilities. Data processed through FC operates within local network regions instead of having to route information through traditional cloud management servers. The processing happens locally at the network edges through fog nodes, which enables both quick responses and low latency that are vital for monitoring systems and emergency response and autonomous vehicles. A healthcare monitoring system makes use of fog nodes to instantly analyze patient data at the edge of the network for early identification of anomalies followed by emergency medical assistance. The implementation of FC in industrial automation helps manage real-time operations of machines for efficient and secure industrial operations [5].

• Prevalent to wireless access: FC has been created to support wireless networks thus making it perfectly suited for new mobile and IoT systems. A variety of wireless communication technologies including mobile cellular gateways along with Wi-Fi access points operate as fog nodes for edge-based connectivity and data processing operations. The integration of wireless access stands vital because there are many situations where traditional wired connections become impossible to utilize. This includes rural areas or situations that occur inside or outside buildings and when using mobile systems. In smart agriculture systems wireless fog nodes operate to grab data from field sensors before conducting spatial assessment of soil situations and meteorological elements and crop vitality at real-time intervals. The capability to employ wireless access throughout FC systems allows its deployment across multiple geographic settings that span smart cities to industrial locations in isolated areas [21].

• Interoperability is the fundamental characteristic of FC involving linking different operational domains through service integration, which allows service federation across multiple platforms. The devices together with applications as well as services, which exist in a FC ecosystem, typically originate from different vendors while using varying protocols. FC platforms need capability to speak with various systems for operation continuity. Wide-area data streaming applications such as video surveillance and telemedicine and smart grid management need this feature intensively. The smart city demonstrates the importance of interoperability among fog nodes from departments such as traffic management and public safety and energy distribution because they need shared data to coordinate city functions. The ability of FC to support interoperability enables the development of unified complex systems which can utilize the advantages of various technology platforms [22].

4 Recent works on minimizing latency in IoT and cloud using FC

This section includes a general review for recent research works to minimize the network usage along with latency and response time in IoT and cloud using FC. Furthermore, the section also includes the various factors affecting the latency in IoT and cloud which prompts delay in the services to end-clients. The section is divided into two sub-sections. The various factors which affect the network and service latency in IoT, and cloud are:

• Propagation delay is caused by a standard number of hops on the route to the destination server. (i.e., the amount of time a packet spends traveling across a wire or fiber; the amount of time depends on the actual distance.) [9].

• Large data transfer to the cloud from healthcare IoT devices.

• There is a ton of information transmission between both the cloud storage and IoT devices. Due to the high measure of information sent by IoT gadgets.

• Network congestion causes router queuing delays, which results in packet and path loss.

• Routers have a built-in processing latency.

• The cloud server's response time.

• Replication of data from one cloud data center to another can cause some service delays.

• An abnormally high burden on the cloud data center.

Therefore, real-world implementation research in this field is possible.

Tripathy et al. [23] proposed a new architecture in a cloud computing environment to integrate wireless body area networks (WBAN) with several modeling techniques for storage and optimization of real-time data. These data were further processed for efficient energy usage and low latency. However, they did not discuss the concept of FC. Kaur et al. [24] introduced the fuzzy inference system discipline along with data mining techniques. They discussed a novel technique for a distributed system. Next they applied fuzzy set rules in their proposed algorithm. The algorithm was based on a breadth-first strategy. Their proposed work supports distributed computing rather than centralized computing. In a similar context, Moustafa [25] used an FC architecture in the IoT-cloud. His work was able to process the IoT device data into meaningful information. The filtered data are further sent to end-users in real-time. Nandyala and Kim [26] presented a three-tier FC architecture for the e-healthcare system. The gap of their research was to overcome the real issues associated with high latency; they focused more on FC modeling techniques. Ijaz et al. [27] discussed the FC for smart devices and networks. The FC here acts as a middleware gateway device. This gateway device works like a computational device with CPU storage and processing capabilities. The gateway device was able to bring cloud features close to IoT devices and end-users. Similarly, Natesha and Guddeti [28] proposed an FC-based gateway to reduce network latency at the edge of networks. However, real world implementation is still lacking.

Tuli et al. [29] highlighted the problem of high latency in the medical healthcare system. They used various cloud-based services for the transmission of e-health data to end-users. However, they did not discuss the concept of FC in the healthcare system. Islam et al. [30] proposed an FC-based intelligent gateway, where fog nodes were used for data storage and distribution to several neighboring fog nodes and end-users. the work was able to reduce the latency in IoT-cloud environment. Similarly, Kaur and Aron [31] proposed an FC-based architecture to highlight the problem of load balancing and data replication among neighboring fog nodes. They pointed out the problem of high latency associated with data replication from one node to another. Li and Wang [32] introduced the concept of green cloud using WBAN for services related to healthcare. But they missed out the issues of high service latency and battery draining in WBAN and end-user devices. Darabkh and Alkhader [33] proposed a hybrid model which is a combination of FC and vehicular ad hoc networks (VANET’s); their proposed model consists of a VANET “hub.” This hub acts like a fog device to deal with large data. This device acts at the edge of networks. Likewise, Aazam and Huh [34] proposed an intelligent gateway that acts as a middleware between body sensors and cloud; their proposed work highlighted the issue of high latency in e-healthcare. The processing, filtering, and mining of healthcare data were done at the gateway. Similarly, Rahmani et al. [35] proposed an enhanced gateway for the healthcare system; they used FC-based technology to distribute the data packets among neighboring fog nodes, which acts as a gateway device. However, their gateway device lacks the decision-making approach.

5 Latency in IoT and cloud reduction techniques

Existing algorithms and techniques for latency reduction in cloud and IoT using FC are discussed in this section.

1. iFogStor: High latency is a big issue for time-critical IoT applications, according to Naas et al. [36]. They proposed the iFogStor method. The method was based on the FC principle. The heuristic technique was employed by iFogStor. It performed far better than the cloud, with latency decreased by more than 86 and 60% when compared to FC alternatives. Data placement issue was defined as a generalized assignment problem (GAP) in iFogStor, and two solutions were proposed: (1) an accurate solution based on integer programming, and (2) a heuristic method based on geographical zoning to speed up the process. Using the geographical zoning heuristic, issues with a high number of fog nodes may be solved quickly and effectively, allowing iFogStor to be runtime and scalable. In the combinatorial optimization literature, GAP is a well-known NP-Hard problem. It entails determining the most cost-effective assignment of n tasks to m agents (for example, n jobs to m processors).

   To limit all out storage and administration latency, iFogStor is a haze mindful runtime approach for putting away IoT information in a mist framework, for example, server farms. iFogStor thinks about the entire progressive plan, as well as the variety of gadgets and latencies between hubs, while putting information. They presented a system for recognizing the generated dataset, along with a strategy for fog storage nodes aimed at minimizing overall service latency during data storage and retrieval. Each piece of data may be stored and accessed with a certain delay, depending on its location and characteristics. The authors formulated the data placement problem as an integer programming model with two proposed solutions. The first was a careful reaction in case of truly huge scope applications, while the second was a gap and vanquish heuristic system to lessen critical thinking time. The idea behind iFogStor is to leverage the heterogeneity and geographic distribution of fog nodes to reduce overall latency in storing and retrieving data. Future improvements will require a more accurate model and architecture to support time-sensitive healthcare IoT applications.

2. FC-cloud fusion: Shukla et al. [37] examined several issues and research problems related to the fog fusion and cloud for latency reduction, energy consumption, allocation of the resources, the optimization, and RAM utilization in IoT. They cast doubt on the expanding disparate devices number.

3. Fog-Radio Access Network (F-RAN): You et al. [38], for lowering IoTs and wireless devices latency, the authors described and recommended a F-RAN architecture. They proposed a latency-driven software algorithm that would enable them to offer services to several end-users with varying resource allocations. A resource may be shared by several users thanks to the algorithm. An F-RAN design that takes use of existing infrastructure, such as small cells and macro base stations, to achieve ultra-low latency via collaborative computing across several F-RAN nodes and near-range communications at the edge. This further defines the tradeoff between communication and computation across a large number of F-RAN nodes. According to their results, F-RAN can supply low latency services using latency-driven cooperative task computing.

4. Live VM migration: Osanaiye et al. [39] introduced an audit of distributed computing research articles, as well as a speculative live VM Migration system to further develop administration latency in IoT-haze cloud design. A notional live methodology for VM movement is proposed to guarantee and give assets and administrations to end-clients. As per the report, the system still cannot seems to be coordinated and sent in the real world.

5. Many-to-one coordinating: Chiti et al. [40] presented and outlined the test of improving cloudlet choice and limiting latency in IoT and haze networks by checking and managing the responsibility. For some-to-one coordinating IoT hubs with cloudlets, it was concocted to rate system. The matching game among IoT and cloudlet was likewise tackled utilizing a calculation.

6. FC security service (FCSS): Wu et al. [41] introduced the requirements for information-centric social networks (ICSN). The ICSN has many needs, including a deployment method, data mobility, low latency, and efficient end-node communication. Computational duties, resources, and intelligence are moved from remote and distant servers to the network's edge when FC is used in ICSN. The services of ICSN are protected by FC-based content-aware filtering. The authors deployed a FC security solution to protect ICSN. The authors did not address the problem of excessive computational delay induced by the fog nodes' enormous number of computational workloads. Large data transmission at the network's edge results in a lot of data traffic and network congestion, as well as a lot of network delay.

7. Cost-effective schema: Yadav et al. [42] discussed issues such as latency, hardware failure, resource limits in FC, and software failure, all of which are considered significant in network function virtualization (NFV). In fog and cloud networks, the authors provided a cost-effective IoT service deployment architecture. The suggested schema assesses the availability of virtual network functions with the potential to enhance software function chaining. Their work is the most cost-effective and scalable, according to them. However, despite the fact that the authors stated that high latency is a key problem for IoT deployment, no effort was done to reduce service function chaining latency via cloud servers and fog nodes.

8. Service popularity-based smart-resources partitioning (SPSRP): Li et al. [43] raised the figuring and asset proficiency issues. On haze hubs, there was an issue with coordination between computational proficiency and asset productivity. Thus, there is a ton of traffic and network clog. The authors fostered a SPSRP (administration fame based savvy assets dividing) method for IoTs and FC. In IoTs and mist servers, their proposed concentrate on exertion diminishes defer time, reaction time, and adaptation to internal failure. The authors, then again, did not resolve the issue of extreme network and handling delay in IoTs.

9. Cloud-based service: Haghgoo et al. [44] discussed the challenges of high latency, huge data transfer, and high data traffic caused by a large number of heterogeneous applications. For healthcare applications, cloud services are unable to fulfill the required QoS. As a result, the authors offered cloud-fog-based services for healthcare applications, as well as a reference architecture. The findings were analyzed for data transmission optimization, latency minimization, and power consumption reduction. Cost efficiency, energy usage, and network latency were improved as a consequence of the study. However, there has not been much research into reducing excessive computational delay in healthcare applications.

10. FC-based system with a central component: Almaiah et al. presented a centric FC-based method for cloud storage [45] to preserve and secure data inside the cloud environment; they highlighted numerous concerns connected to data security, data transmission latency, and cloud privacy. In cloud servers, the privacy of users' data is paramount. To safeguard the data from unwanted access and harmful assaults, a XoR combination was utilized. In terms of data packet processing time, the findings were confirmed. To identify data change with the highest likelihood, the scientists employed a novel approach based on a hash algorithm.

11. Software-defined network (SDN): Lakhan et al. [46] proposed an SDN for IIoTs: A technique for task execution based on task priority using a computing mode system. Real-time performance is accomplished with FC when the task priority is used. They were able to minimize IoTs latency, fog, end-users, and cloud using their suggested strategy. However, there are no computing services for mobile devices in the study effort.

6 Minimizing packet loss and data traffic

The problem of excessive packet loss and data traffic in the cloud and IoT is highlighted in this section. An itemized conversation and examination of flow research work are embraced to packet parcel loss and information traffic utilizing FC and regular methods.

1. Offloading calculation: Cao and Cai [60] laid out a system for asset enhancement and portion. Dividing the information parcels into smidgens may be utilized to convey assets. These bits are then conveyed to different hubs in the network. The work is scattered to other sub mists called cloudlets in view of the size of the errand and the quantity of assignments holding up in a line, a cycle known as figure offloading. This procedure will aid in node load balancing. This was the best choice for unloading. The suggested technique determines the resource capabilities of nodes with the goal of reducing total data traffic rate. They presented about how cloud computing may be used to offload computation for numerous users in a cloudlet context. The user has the option of offloading the data based on his or her own set of criteria and information. The scientists also presented a machine-learning system for mobile devices and apps that would cut energy use, bandwidth consumption, and network utilization.

2. Cloudlet: A cloudlet was suggested by Cavalcante et al. [61] to reduce the amount of data transmission between the edge device and the cloud. Various virtual machines make up the cloudlet. The devices were built with the intention of moving from one cloudlet to the next. According to the authors, this option allows the program to run quicker. The authors, however, confronted difficulties as a result of the application's long reaction time.

3. FC framework: Liu et al. [62] presented an FC framework. They presented a latency-reducing design and explored many problems and optimization options. They also spoke about resource allocation approaches for reducing load and network traffic while employing FC. However, there is no real-world application of the study findings.

4. Asset portion: Name et al. introduced and proposed a strategy [63] to deal with the asset allotment issue in IoT-mist cloud design. The method joins a portable IPV6 handover system and booking calculations to limit network use, information traffic, response time, latency in IoTs, and much more. However, their study does not include any work on reducing the amount of RAM used by processes.

5. IoT and cloud interconnection: Al-Fuqaha et al. [64] utilized FC to lead a profound and complete examination of the connections between distributed computing, edge figuring, and IoTs. They likewise featured on how current and impending advances could lessen latency and information traffic between cloud, end clients, and IoT gadgets. The dependability of crisis reaction application administrations, then again, is a main issue, bringing about huge deferrals and information misfortune.

6. Communication that is cooperative: Masri et al. [65] depicted how to involve FC to lessen network traffic and latency in an IoT-cloud situation. They contrived a methodology for haze hubs to cooperatively convey. In IoT-haze cloud framework, this assists with lessening information traffic and start to finish latency. Their review needs true application to check the worldview they offer.

7. FC design and computation offloading: Meng et al. [66] proposed a mixture system to lessen network traffic and energy utilization in IoTs using FC. Naha et al. [67] investigated the FC plan for limiting network traffic and latency in IoTs. They likewise fostered a scientific categorization for FC with regards to IoT.

8. IoT system: Yousefpour et al. introduced the IoT-mist cloud structure [68]. Also, they gave a logical way to deal with lessening IoT administration latency. The authors proceeded to make sense of how the FC method might be utilized to apply the idea of stretching out cloud administrations to the edge of networks to increment framework execution and asset proficiency. The proposed design limits network transfer speed use as well as information traffic.

9. FC review and examination: Mukherjee et al. [69] distributed a total report on FC to lessen network traffic and latency in IoT. Also, the authors featured network applications, research issues, and FC establishments according to IoT and cloud. Besides, FC can possibly further develop IoT gadget execution and proficiency. Also, Mouradian et al. [70] led a careful assessment of the FC research issues. The study incorporates an FC scheme for latency reduction and a network traffic management algorithm. However, the findings of the study have not yet been applied in a real-world implementation.

10. Virtual haze system: A virtual mist structure was introduced by Li et al. [71]. To depict the virtual haze in the IoT framework, a NFV was proposed. This virtual methodology in FC had the option to lessen jitter, latency, and information traffic. In any case, in the proposed task, asset use is a vital issue.

11. FogBus: Tuli et al. [72] proposed a FC scheme for latency reduction, along with a network traffic management algorithm. However, the study's findings have not yet been validated through real-world implementation. Their proposed arrangement utilizes a blockchain-based innovation to lessen weighty information traffic while likewise shielding private IoT information from outside interlopers and programmers. To protect sensitive activities involving IoT data, the system employs multiple encryption techniques. Additionally, the proposed architecture enhances the integration of the IoT–Fog (Edge)–Cloud framework. FogBus gives stage skeptic execution and connection interfaces for IoT applications and figure cases. It assists designers with making applications, yet it likewise helps buyers in running numerous applications immediately and specialist co-ops in dealing with their assets.

7 Techniques to achieve QoS requirement for IoT

In the medical care industry, only a couple of uses need ongoing information. In such conditions, FC-based innovation is vital for meeting the medical care IoT's QoS needs. The IoT device might send information continuously utilizing this method. By lessening information transmission, FC makes communication between haze hubs and end gadgets simpler. Bluetooth and Zigbee are utilized to impart among these gadgets [73]. Painted health data (PHD) and important bodily functions are recorded by FC at the network's edge. These clinical devices are normally used in far off areas. The FC fulfills the QoS rules for medical care time-touchy applications when contrasted with the cloud. FC is important for a scattered neighborhood local area network. Cloud, then again, is more unified and utilizes a wide area network (WAN). The FC contraption is a savvy gadget. Due to its nearness to IoT gadgets, haze gadgets might execute an assortment of time-touchy medical care applications [74]. The broad sending of savvy edge gadgets and applications that need ongoing information handling has without a doubt required the expansion of distributed computing to the edge, otherwise called mist or edge figuring. FC is planned to work in pair with distributed computing, considering a new, more extravagant design that can take utilization of and incorporate both haze and cloud assets.

1. FogPlan: Yousefpour et al. [75] introduced a system called FogPlan to fulfill the QoS rule for low latency. The system was based on a unique help that gave mist hub applications. In their proposed study, they utilized an eager calculation. The convention design for administrations used between haze hubs is absent from the proposed work. There are also no solutions available for managing traffic between fog nodes that rely on learning. FogPlan is a platform designed to dynamically provision fog services while maintaining Quality of Service (QoS) through quality of dynamic fog service provision (QDFSP). QDFSP focuses on the adaptive deployment of application services on fog nodes, or the reallocation of currently deployed services, in order to meet the low latency and QoS requirements of applications, while simultaneously minimizing costs.

2. FogTorch: Brogi and Forti [76] proposed a worldview for latency and transfer speed improvement in IoT and cloud in view of a Java program named FogTorch. The model follows IoT QoS necessities. The model portrays functional fundamental elements of accessible framework (latency and transfer speed), programming part and thing connections, and business rules. The calculations proposed here handle these issues by adaptively figuring out where a part ought to be sent along the Cloud-to-Things continuum. To limit the hunt space and observe a solitary qualified sending, the authors utilized a pre-handling and backtracking method utilizing a heuristic system. The proposed model and methods are executed in a proof-of-idea Java program called FogTorch to show their innovative suitability. FogTorch takes the framework and application details, as well as the pertinent things restricting and sending strategy, and creates reasonable arrangement plans. The proposed model might be utilized during runtime, planning, and sending. The methodology deliberately sticks to no norm for characterizing equipment and programming abilities. Programming abilities included the accessible operating systems (Linux, Windows, and so on) and introduced stages or systems, while equipment details stressed both consumable (e.g., RAM, storage) and non-consumable assets (e.g., design, CPU centers) (.NET, JDK, Hadoop MapReduce, and so on.). Besides, the authors laid out a haze framework in their model, yet they did not zero in on the model's constancy, mist hub sending timing, or cost.

3. E-medical care needs examination and correlation: Skorin-Kapov and Matijasevic investigated the QoS necessities for e-medical care [77]. This remembers necessities for ongoing information transmission for medical care IoT and telemedicine. Negligible full circle time deferral, or least help delay, is a rule for IoT gadgets and telemedicine tasks. The authors led a careful examination and correlation of e-medical care administration QoS necessities.

4. FC and savvy medical care: The importance of IoTs in shrewd medical care applications was featured by Baker et al. [78]. Different advances, troubles, and potential outcomes connected with medical care IoT, e-medical care, and telemedicine are examined. The authors proceeded to investigate how FC might assist with lessening latency and accomplish QoS necessities for time-basic applications in savvy medical care. They introduced a unique system for following patient wellbeing. The authors then went through the various wearable gadgets that might be utilized to screen and record patient crucial signs. In their proposed framework, they applied an AI method.

5. Healthfog: Tuli et al. [29] proposed a structure for a savvy medical care framework that coordinates IoT with FC to identify patient heart issues continuously utilizing profound learning-based approaches. The proposed design had the option to satisfy the medical care IoT QoS necessity. The system was made in view of latency-touchy applications, for example, clinical checking and flight control. The IoT in medical care makes a huge amount of information known as Big information. This information requires a huge estimation, after which the information is moved to data sets and from data sets to cloud server farms, bringing about a decline in framework execution. Thus, the proposed haze empowered cloud design fulfills the QoS for medical care IoT regarding power utilization, jitter, and coronary illness analytic forecast precision.

8 FC performance tools

The various simulation tools for evaluating the performance of the FC technique are covered in this section. This section also includes a full comparison of current tools. Framework (i.e., gadgets and networks), stage (which incorporates assets, administrations, and their organization), and applications are immensely significant parts of FC frameworks (which execute in the mist and have specific necessities). Resource management, scalability, and flexibility are all essential aspects of FC modeling and simulation [79]. FC, however, represents a complex environment in which users both consume services and execute their applications [80]. The allocation of services or resources depends on the roles of the participating nodes, whether as providers or consumers.

The FC worldview fits a wide assortment of utilizations with low latency necessities, time awareness, and assumptions for openness or proficiency from portable clients. Medical care, savvy environmental factors, augmented reality, intelligent transportation systems, public security, smart grid, and Industry 4.0 are only a couple of the areas where these applications might be found. To help the displaying of numerous sorts of uses, recreation devices incorporate choices for physical or virtual assets, network design, control systems, and information of the board [79]. FC is the most recent cloud computing expansion [80]. Although there is a large variety of cloud computing simulation tools, they cannot be utilized as-is for FC research; hence, they have been customized to fit the new demands. Simultaneously, new simulation tools, particularly for fog, have been suggested and created. This section does a survey study of all such instruments in the FC region.

The available simulation tools are summarized in Table 1. The first column refers to the issue that the different simulators are attempting to solve. The second column contains the metrics supplied by each simulator, which emphasizes the simulator's major goals. The list of simulators is included in the third column. iFogSim seems to be the most popular tool among researchers based on these data. The majority of simulators are designed to help with a particular FC issue.

Table 2 presents complete details about different FC simulators that include their implementation platforms with fundamental metrics alongside their objectives for use. The field of FC needs crucial information about simulation tools to allow researchers and practitioners to make appropriate tool selections according to their distinct requirements. The table functions as a key reference by identifying the specific focus areas among different simulators, which include QoS, energy consumption, latency, and resource management capabilities. The decision-making process for FC system design becomes simpler through this tool especially regarding healthcare applications that need low latency and resource-efficient management. Also, Table 2 demonstrates how FC simulators possess diverse specializations, which align with the complexity of FC difficulties. Both iFogSim and MyiFogSim focus on resource management and latency since healthcare applications need real-time responsiveness yet other tools like FogTorch and FogNetSim++ tackle general aspects including reliability and security and power efficiency. Researchers can focus on solving particular fog computing problems due to the availability of various simulators, which tackle both performance optimization and energy-efficient operations. The table integration allows users to choose suitable tools and exposes development opportunities which guide the creation of advanced fog computing simulators for emerging application requirements. A technical analysis of the simulation tools is presented in Table 2 based on their capabilities. As observed, current simulators address only a portion of the simulation requirements in fog computing (FC). While most of them support multiple devices and components for fog infrastructure development, they vary in terms of the features and interactions they provide. To evaluate the suitability and effectiveness of each tool in addressing specific challenges in FC and IoT, a comprehensive assessment of existing simulation platforms has been conducted.

9 Identified research gaps

The most current study does not provide any recommendations for reducing latency. This is due to the fact that current data analytic techniques are built to handle enormous amounts of data but not real-time data processing and dispatching. Having millions of objects generate data and then sending it all to the cloud is neither scalable nor ideal for real-time decision making. Time-dependent IoT environments together with instantaneous requirements continue to grow at a rapid pace. Due to the processing ability of edge devices, it becomes necessary to implement a real-time fall detection system. The detection workload separation between edge devices linked to users and cloud servers enables this system to use the FC paradigm for analytics distribution throughout the network. FC paradigm originated from cisco as an answer to provide QoS requirements for IoT applications. Different authors described how FC fulfills the (network speed and system response time requirements) that IoT technology demands. FC ensures all latency reduction factors to make it an excellent choice for real-time data transmission between IoT devices and cloud systems. The main barrier resides in developing edge device resource selection protocols to determine which analytics applications will be delivered per device as a way to reduce delay and increase data speed. The assessment platform needs development for measuring how resource management rules execute on both IoT and FC infrastructure through repeatable testing methods. The key focus of this project exists in advancing IoT-based systems through the introduction of both smart FC architecture and paradigm. The IoT device performance and efficiency benefits from the implementation of forwarding capability algorithms. This framework ends cloud services to form a wider system that enables connected objects to communicate via wireless radio waves. Analysis and comparison of methodologies proved that researchers failed to dedicate sufficient work to reduce complete latency, which includes compute delay together with communication latency and network latency between IoTs and cloud. A new technique needs development to lower substantial latency amounts required for time-sensitive healthcare IoT applications.

The research of latency optimization for H-IoT systems in FC requires additional work to resolve multiple outstanding gaps before establishing improved solutions. The following essential deficiencies exist in latency optimization approaches for H-IoT platforms in FC context:

• Multiple strategies for latency reduction (such as architectural improvements and communication protocols and load balancing) remain unconnected through comprehensive overall frameworks within many existing approaches despite their focus on specific aspects (such as task scheduling or resource allocation).

• The analysis of latency optimization methods lacks confirmation in practical medical centers through extensive testing. Proposed solutions that scientists test in artificial conditions might fail to demonstrate accurate results in healthcare facilities with their real-world complexities and unpredictable scenarios.

• The process of enhancing performance through latency optimization leads to increased energy usage during execution. The H-IoT community requires additional research to develop strategies, which reduce latency yet maintain high energy efficiency to achieve both performance speed and sustainability.

• The performance of latency optimization methods remains unstable when healthcare systems grow because most solutions were designed primarily for small environments. The processing of expanding H-IoT systems presents a major problem because it requires effective solutions to handle increasing devices and data streams.

• The enhancement methods for reducing latency need to ensure both security requirements and privacy standards since they should protect medically sensitive patient information. The integration between fast solution deployment systems and secure infrastructure remains a problem that researchers need to address.

• The healthcare environment operates in a dynamic state because it deals with fluctuating demands together with shifting operational conditions. A gap exists in current approaches because they fail to support immediate adaptation according to live network changes and device availability and data flow patterns.

• Different FC platforms and H-IoT devices work less effectively when standardized protocols and interoperability between these systems are absent. Standardization along with universal guidelines remains essential to achieve perfect performance and system integration.

10 Conclusion

A review on latency reduction in healthcare IoT and cloud computing has been conducted. The limitations of the current systems and technologies have been identified, revealing that several performance criteria remain unaddressed. Additionally, processing delays, network latency, and communication latency were found to be significant and often impractical for real-world applications. As a result, various latency reduction approaches and technologies have been explored. By reviewing systems, tools, challenges, methodologies, and technologies, this study aims to bridge the existing gap. This study delves into the systems and technologies used in IoT and cloud computing to reduce latency. Furthermore, the findings are expected to serve as a foundation for future research. This study intentionally selected and thoroughly analyzed relevant articles for the review, synthesizing data from these articles to address a set of research questions outlined in this study, which guided this investigation. The objective of this study is to provide an in-depth examination of the latency issue in IoT and cloud computing, as well as to propose a novel solution and compare it with other latency reduction strategies currently in use. The goal is to encourage future research toward developing a latency-aware model for healthcare IoT. The contributions of this study can be summarized as follows:

1. Classification of technologies and approaches: A systematic categorization of the technologies and approaches discussed in the literature.

2. Quantitative analysis: Evaluation of the volume of distributed works analyzed and search trends in Google Scholar to identify patterns in latency reduction techniques and technologies.

3. Review of latency reduction strategies: A comprehensive review of various latency reduction strategies and technologies.

4. Identification of research gaps: Recognition of research needs in the areas of IoT, FC, and cloud latency reduction.

5. Addressing limitations: Highlighting the limitations of existing studies and unresolved challenges related to latency requirements for time-sensitive applications.

From this review, researchers and industry professionals will gain a clearer understanding of IoT and cloud computing requirements for time-sensitive applications, along with a better grasp of latency reduction approaches and technologies. The investigation presents different techniques which use FC to decrease healthcare IoT system latencies. The time it takes to process and respond to information remains a decisive aspect of healthcare applications when they need to provide effective patient care through remote monitoring and emergency situations and telemedicine systems. Cloud computing proves inappropriate for time-sensitive healthcare applications because it presents two major limitations through high latency together with bandwidth constraints that impact its performance. The authors introduce FC as a potential solution because it positions computing and storage near network edges to solve current limitations. The analysis divides latency optimization strategies into four main sections which encompass resource distribution along with task delegation and information compression and edge storage functions. Researchers analyze the approaches through two sets of metrics including effectiveness and implementation complexity as well as suitability levels across healthcare conditions. The study explores how machine learning alongside artificial intelligence enhances real-time latency prediction and management. Predictive analytics models allow organizations to distribute resources in advance when they anticipate upcoming demands. The authors recognize that hybrid fog-cloud model selection plays an essential role when designing network architecture since it allows operators to maintain system efficiency while reducing latency times. The research identifies security and privacy requirements for healthcare IoT before discussing ways to protect patient confidentiality while optimizing latency. This review suggests several directions for future research such as standardizing protocols and examining 5G networks and building better facilities for validating these methods in genuine healthcare settings. This study presents extensive details about recent approaches for healthcare IoT latency reduction together with an emphasis on how FC supports quick and effective medical care systems.