Network latency, the time it takes for data to travel between Internet of Things (IoT) devices and cloud-based servers, is a critical factor in the performance and responsiveness of IoT applications. Minimizing latency ensures real-time data processing, timely decision-making, and system efficiency. Several strategies can be employed to optimize network latency in cloud-based IoT systems [1]. Edge computing is one of the most effective approaches to reduce latency [2]. By moving data processing closer to the IoT devices, edge computing eliminates the need for data to travel long distances to the cloud. This reduces network congestion and minimizes transmission delays.

Additionally, edge computing can enable local decision-making, allowing faster responses to real-time events without relying on cloud-based processing [3]. Optimizing network infrastructure is another crucial aspect of minimizing latency. Upgrading routers, switches, and other network components to support high-performance data transmission can significantly improve network throughput and reduce latency. Implementing Quality of Service (QOS) mechanisms can also prioritize IoT traffic, ensuring that critical data packets are delivered with minimal delay. Choosing the right communication protocols can also impact network latency. Protocols for low-latency communication, such as MQTT or COAP, are well-suited for IoT applications [4]. These protocols use efficient message formats and minimize overhead, resulting in faster data transmission. Cloud providers offer various network options, including regional and global data centers. Selecting the appropriate data center location can significantly reduce latency for IoT devices in specific geographic regions.

Additionally, leveraging Content Delivery Networks (CDN) can cache data closer to IoT devices, reducing the need for data fetched from distant servers. Finally, optimizing application design and data management practices can also reduce latency. Organizations can reduce network traffic and improve latency by minimizing the amount of data transmitted between IoT devices and the cloud and using efficient data compression techniques.

The motivation behind our work stems from the pressing need to reduce network latency in cloud-based IoT systems and help with its early detection and prevention through various networking models. The emergent IoT is thought to be the next generation of the internet, in which billions of things are interconnected. Examples are sensors, actuators, mobile phones, and cars communicating with each other to perform service objectives. Cloud computing is a new paradigm that enables users to elastically utilize a shared pool of cloud resources (e.g., processors, storage, applications, services) in an on-demand fashion.

The emergent IoT is thought to be the next generation of the internet, in which billions of things are interconnected. Examples of these things are sensors, actuators, mobile phones, and cars communicating with each other to perform service objectives. Cloud computing is a new paradigm that enables users to elastically utilize a shared pool of cloud resources (e.g., processors, storage, applications, services) in an on-demand fashion. Recently, driven by the potential of complementing the ubiquitous data-gathering abilities of IoT devices with the powerful data storage and data processing capabilities of the cloud, the integration of the cloud and the IoT is attracting rising attention from academia and industry [5].

The data (alarm, security, climate, and entertainment) gathered by sensors are transmitted first to the gateway. The associate editor coordinating the review of this manuscript and approving it for publication was Adnan M. Abu-Mahfouz. Then, the received sensory data is transmitted to the cloud. Eventually, the cloud stores, analyzes, processes, and transmits the sensed data to the users on demand. During the entire data transmission process, if the data transmission from the sensor nodes to the cloud is unsuccessful, data are re-transmitted until successfully delivered. For this cloud-based IoT prototype, the IoT acts as the cloud's data source, while users are the cloud's data requesters. The users can access the needed sensory data from the cloud whenever and wherever there is a network connection. In these potential applications of cloud-based IoT integration, such as smart buildings of smart cities [6], a number of them require the IoT will reliably offer sensory data to the cloud based on users' requests. In general, sensors' limited battery power will be depleted by performing data sensing, processing, and transmission after a specific period, as they are usually supplied with non-rechargeable batteries, and their replacement may also be unpractical [7]. Several approaches have been evolved VOLUME 7, 2019 2169-3536 2019 IEEE. Translations and content mining are permitted for academic research only. Personal use is also allowed, but republication/redistribution requires IEEE permission. However, approaches to reduce power consumption contrarily impact the network's reliability. An example of this approach is applied when part of the network works whilst other parts sleep. This approach is excellent for power consumption but not reliability because part of the network may be inaccessible due to an IoT node sleeping. Another example of this approach is multiple paths between a specific IoT node and the gateway (Fig. 2).

In contrast to the previous example, this method is excellent for reliability but not for power consumption because it will use more than one route, which means more IoT nodes to transmit the same packet. Therefore, assessing the reliability of IoT by considering traffic power consumption is important. In this work, we have considered that IoT networks can fail in two points: links due to traffic congestion or interference and sensor nodes due to diminishing their energy. This paper proposes four models for achieving two goals; the energy efficiency of cloud-based IoT considering the reliability level. For instance, we used the following approaches to reduce total traffic power consumption. First, each model aims to minimize the total transmitted power by selecting the IoT device with the lowest energy per bit and idle power. Second, using the data compression technique reduces the amount of data transmitted. Thirdly, interference cancellation will reduce the re-transmission of the data lost due to interference. To achieve the reliability objective, we proposed two approaches; the first is to select reliable links with a 99% reliability level. The second is to choose a standby link as an alternative to the link failing. Finally, each proposed scheme has two optimization objectives: energy efficiency and reliability. The main contributions of this paper are summarized as follows:

I. Virtualize cloud-based IoT network using the MILP model.

II. Minimize the total traffic power of the cloud-based IoT network through the MILP optimization model.

Minimize interference.

III. Achieving reliability in the cloud-based IoT network.

IV. Distributing traffic through the gateways to avoid traffic congestion to the cloud in the IoT network. Handle more traffic demands by using the data compression technique.

V. Jointly investigate the issues regarding energy efficiency and reliability from the viewpoint of cloud-based IoT integration.

This paper proposed four models related to the evaluation of cloud-based IoT networks. It considers the mote energy level as the main factor of failures of WSN nodes. It uses the routing algorithm to define the paths between different WSN regions and the sink node, and it automatically generates reliability models considering the elements above. This paper further proposes four schemes consisting of a Standby Routes Selection Scheme (SBRS), a Desired Reliability Level Scheme (DRLS), a Reliability-Based Sub-Channel (RBS) scheme, and a Reliability-Based Data Compression (RBDS) scheme aimed at improving the reliability of IoT networks and reducing total transmitted power. Specifically, an SBRS selectively chooses standby routes to overcome node failure problems and reduce transmission power. In addition, a DRLS is used when a specific reliability level is needed to guarantee the link's reliability while minimizing transmission power. Furthermore, due to its high reliability, an RBS uses sub-channels to mitigate interference and reduce overhead on links that are utilized by several IoT devices. Finally, an RBDS uses a sequential lossless entropy compression (S-LEC) data compression algorithm to overcome the capacity limits of the links and reduce transmission power. In the rest of this paper, Section II introduces the related work, Section III describes the background to this research, and Section IV presents the cloud-based IoT integration system model. Section V introduces the network optimization model of cloud-based IoT. Section VI presents our model objectives by introducing SBRS, DRLS, RBS, and RBDS, and Section VII evaluates the model results. In Section VIII, we state our conclusions regarding the research.

Edge computing and cloud computing have become essential in enhancing the performance of IoT systems. By integrating the computational capabilities of the cloud and the low-latency benefits of edge devices, edge-cloud systems offer a robust solution for IoT deployments that require real-time data processing and low energy consumption [8].

After reviewing works improving IoT systems' performance, we found that most solutions propose improving one part of the system at a time: the network, the processing, or data quality. Per contra, reducing the sent data enhances the network's and the server's responsiveness and saves the IoT device's energy and resources. However, propositions in data transmission reduction are limited since most rely on classifiers, compression, or prediction methods. Those methods take a long time to train and add heavy processing on the resource-limited object, and on top of that, they are not one hundred percent error-free. Also, data transmission solutions focus on one IoT system, generally WSN containing temperature, humidity, and luminosity sensors, since they generate numeric data that can be predicted easily. Moreover, for all these solutions, no generalization is proposed. Therefore, our work is characterized by proposing a generalized (that can be projected on different IoT systems) and formalized (as we define mathematically all the functioning of the model) solution to reduce data offloaded from a connected device to the cloud. This improves the system's performance in several aspects.

The optimization of network latency in cloud-based IoT systems is crucial to ensuring real-time data processing, improved user experience, and enhanced system performance. Techniques like edge computing, efficient data routing, protocol optimization, and load balancing can significantly reduce latency. Edge computing minimizes the distance between data generation and processing by offloading tasks to local servers, reducing round-trip times to the cloud. Adaptive data compression and caching further lower transmission delays, while advanced protocols like QUIC optimize data flow. Additionally, leveraging AI and ML for predictive analytics can preemptively address network bottlenecks, resulting in a more responsive and scalable IoT ecosystem. Ultimately, these strategies collectively enhance cloud-based IoT networks' overall efficiency, reliability, and speed.

This research received no external funding.

The data used and analyzed during the current study are available from the corresponding author upon reasonable request.

The authors declare no conflict of interest regarding the publication of the paper.