Cloud Computing Paradigms: 5G, Cyber-Physical Systems, Spatial Cloud, and IoHT

5G Network Technology and Evolution

5G is the fifth-generation mobile network, representing a new global wireless standard that follows the 1G, 2G, 3G, and 4G networks. It is designed to enable a new kind of network capable of connecting virtually everyone and everything together, including machines, objects, and devices. This technology is intended to deliver significantly higher peak data speeds in the multi-Gbps range, ultra-low latency, enhanced reliability, massive network capacity, and increased availability. The ultimate goal is to provide a more uniform user experience to a larger number of users, empowering new user experiences and connecting diverse industries through improved efficiency and performance.

Comparative Evolution of Mobile Generations

The evolution of mobile networks began in the 1980s with the first generation (1G), which delivered analog voice. The second generation (2G) emerged in the early 1990s, introducing digital voice and technologies like Code Division Multiple Access (CDMA). By the early 2000s, the third generation (3G) brought mobile data, exemplified by CDMA2000. The fourth generation (4G LTE) arrived in the 2010s, ushering in the era of mobile broadband. 5G, deployment of which began around 2020, is a unified and more capable air interface designed with extended capacity to enable next-generation services and deployment models.

In terms of theoretical performance, 1G provided basic voice, while 2G offered download speeds of approximately $2\,\text{kbit/s}$ with a latency of $629\,\text{ms}$ . 3G increased speeds to $384\,\text{kbit/s}$ with $212\,\text{ms}$ latency. 4G LTE reached speeds of $1\,\text{Gbit/s}$ with latency ranging from $60\,\text{ms}$ to $98\,\text{ms}$ . In contrast, 5G aims for theoretical download speeds of $10\,\text{Gbit/s}$ and a latency of less than $1\,\text{ms}$ . Features of 5G compared to 4G include a jump from $100,000$ to $1,000,000$ connections per $km^2$ , and a shift from monolithic, centralized physical architectures to virtual, dynamic, and distributed ones. Data traffic is projected to grow from $7.2\,\text{Exabytes/Month}$ in the 4G era to $50\,\text{Exabytes/Month}$ in the 5G era.

Primary Use Cases and Features of 5G

5G is designed for forward compatibility, allowing it to flexibly support future services. It focuses on three main types of connected services. First, Enhanced Mobile Broadband (eMBB) provides faster and more uniform data rates for smartphones and immersive experiences like Virtual Reality (VR) and Augmented Reality (AR) with a lower cost-per-bit. Second, Ultra-reliable and Low-Latency Communications (URLLC), also known as mission-critical communications, enables services requiring high reliability and low latency, such as remote control of critical infrastructure, autonomous vehicles, and remote medical procedures or telehealth. Third, Massive machine-type communications (mMTC) or Massive IoT aims to connect a huge number of embedded sensors (e.g., in smart cities, asset tracking, and smart agriculture) by scaling down data rates and power consumption for low-cost connectivity.

Convergence of 5G and Cloud Computing

5G is considered a perfect companion to cloud computing due to its distributed nature and the diversity of compute and storage capabilities it supports. On-premises and edge data centers are increasingly bridging the gap between resource-constrained devices and distant cloud centers, necessitating heterogeneous and distributed computing architectures. This evolution requires service providers to offer end-to-end orchestration and automated service layer agreements. Network as a Platform (NaaP) becomes a reality where service orchestration allows industrial applications to influence traffic routing and select specific Quality of Service (QoS).

There are two key aspects of this relationship: the development of cloud computing through edge, mobile edge, and fog computing to meet 5G needs, and the "cloudification" of 5G itself through network softwarization, Network Function Virtualization (NFV), and Software Defined Networking (SDN). This convergence leads to hyper-connectivity, where computing, cloud, and IoT systems interact seamlessly.

Edge and Mobile Cloud Computing in 5G

Edge computing is an emerging technology that facilitates 5G by bringing cloud capabilities closer to end users or User Equipment (UE), overcoming high latency issues inherent in traditional centralized clouds. This is particularly vital for interactive, computationally-intensive applications like AR and VR. 5G must handle massive data from mobile/IoT devices while meeting stringent QoS requirements and supporting heterogeneous environments for interoperability. Applications of edge computing in 5G include healthcare, entertainment, the tactile internet, and intelligent transportation systems.

Mobile Cloud Computing (MCC) involves sharing resources for mobile applications to improve reliability through cloud backups. By offloading intensive processing from mobile devices to the cloud via base stations or satellites, device resource consumption is minimized. This system involves a complex interaction between mobile users, network operators (ISPs), and data center owners or cloud service providers using components like Base Transceiver Stations (BTS), Access Points, and Cloud Controllers.

Cyber-Physical Systems (CPS) Fundamentals

A Cyber-Physical System (CPS) is an orchestration of computers and physical systems where embedded computers monitor and control physical processes through feedback loops. The term was coined by Helen Gill at the National Science Foundation (NSF), USA, in 2006. CPS represents the intersection of the physical and cyber worlds, merging engineering models (mechanical, electrical, biomedical, etc.) with computer science methods. CPS components include physical components and software that are deeply intertwined, operating on different spatial and temporal scales and exhibiting multiple behavioral modalities. Key elements include Cyber, Physical, Computation, Dynamics, Communication, Security, and Safety.

Applications for CPS are vast, including precision tasks like robotic surgery and nano-level manufacturing, operations in dangerous environments like deep-sea exploration, and coordination efforts like air traffic control. Advanced SNSS (Smart Networked Systems and Societies) use sensors and actuators to link computational systems to the world, viewing computing as a physical act where reality is monitored in cyberspace to affect the physical environment.

Cyber-Physical Cloud Computing (CPCC)

The Cyber-Physical Cloud Computing (CPCC) architectural framework is defined as a system environment that can rapidly build, modify, and provision CPS composed of cloud-based sensors, processing, control, and data services. CPCC benefits include efficient resource use, modular composition, rapid development, scalability, and a reliable, resilient architecture. A typical framework includes an application plane, a cloud plane, and an edge plane where local CPSS (Cyber-Physical-Social Services) integrate distributed decompositions for smart cities, homes, and factories.

Spatial Cloud Computing and Analysis

Spatial or geospatial data describes objects or events with a location on or near the Earth's surface, combining location information (coordinates), attribute information, and temporal information. Spatial analysis involves solving location-oriented problems to understand regional characteristics and relationships. Typical challenges include handling voluminous spatio-temporal data, integrating heterogeneous sources, and providing real-time IT resources for emergency responses.

Spatial Cloud Computing (SCC) refers to a cloud paradigm driven by geospatial sciences and optimized by spatiotemporal principles. It provides an elastic platform for phenomena simulations, analytical visualization, and decision support. Advantages of SCC include ease of use, scalability (no need for physical purchase), cost-optimization based on usage, and high reliability using enterprise-grade hardware. A typical SCC architecture involves OGC (Open Geospatial Consortium) clients, spatial databases like Oracle Spatial, PostgreSCL, MongoDB, or MySQL, and geo-file storage for vector and raster data published via WFS, WMS, and WPS instances.

Trajectory Cloud (Traj-Cloud) for Urban Dynamics

A spatial trajectory is a trace of a moving object represented as chronologically ordered points <(lat_1, lon_1), t_1>, <(lat_2, lon_2), t_2>, \dots. Semantic trajectories include additional information like activity performed at stay-points. Traj-Cloud is a framework designed to analyze urban dynamics using mobility traces to improve Intelligent Transportation Systems (ITS). It minimizes service-waiting times for services like food delivery or medical emergencies.

Traj-Cloud services include:

Trajectory Data Indexing Service (TS1): Inputs GPS traces and outputs spatio-temporal indices using Google BigQuery and Cloud SQL.
Trajectory Map-matching Service (TS2): Projects GPS traces onto road networks using MapReduce on Google Compute Engine.
Trajectory Query Service (TS3): Processes point or range queries to retrieve trajectory information.

Internet of Health Things (IoHT) and Fog Computing

Fog computing brings the cloud closer to data-producing sensors by using routers, servers, and switches as fog nodes. In an IoHT context, fog computing addresses cloud limitations such as latency and bandwidth requirements while overcoming IoT device limitations in processing and power. A Cloud-Fog-Edge-IoT hierarchy organizes processing: the Edge layer handles large-volume real-time data at the source, the Fog layer performs local data analysis and reduction, and the Cloud layer manages big data processing and business intelligence.

The Cloud-Fog-Edge-IoHT model aims to reduce latency and costs. Using the iFogSim simulator, hierarchical network topologies are modeled with levels: Level 0 (Cloud), Level 1 (ISP), Level 2 (Area Gateway/Fog), and Level 3 (Mobile/IoT). Performance is evaluated based on average latency, network usage, cost of execution, and energy consumption. For example, in a fog-based model, modules like Data Filtering and Processing are placed at the Area Gateway, whereas in a cloud-based model, they are moved to the cloud, significantly increasing latency and network usage.

Hardware Implementation and Cardiac Prediction

Actual implementation of the IoHT model uses Raspberry Pi as fog devices and AWS as the cloud. Sensors like the ADXL345 accelerometer and NodeMCU ESP8266 boards are used for data collection. Activity detection is performed using an accelerometer where data extraction follows the formula $A = \sqrt{x^x + y^y + z^z}$ . After a 5-point smoothing process to reduce noise, features like maximum/minimum amplitude, mean, and standard deviation are extracted and passed to a K-Nearest Neighbor (KNN) classifier. Feature normalization is calculated as $Y = \frac{x - \text{min}}{\text{max} - \text{min}}$ .

A case study for cardiac attack prediction uses the logic: $p = \text{(BPM/f)}$ $s = \text{(systolic measurement)/f}$ $d = \text{(diastolic measurement)/f}$ If $(p \geq 170)$ and $(s \geq 180)$ and $(d \geq 120)$ , then HeartAttackAlarm is set to true. (Note: This is for demonstration purposes and has no clinical significance).

Dew Computing in Healthcare

Dew computing is an on-premises software-hardware paradigm where the local computer provides functionality independent of the cloud while remaining collaborative. In a Dew-based IoHT framework, database synchronization occurs when internet connectivity is restored. This provides a robust alternative when the cloud is inaccessible. Comparing models, a Cloud-Fog-Edge-Dew configuration offers optimal on-premise resource utilization, high uptime despite low internet connectivity, and low latency/bandwidth requirements, although it possesses limited local processing power compared to the centralized cloud.