Secure Smart Healthcare Monitoring in Industrial Internet of Things (IIoT) Ecosystem with Cosine Function Hybrid Chaotic Map Encryption

The technological progression is raised as a hybrid ecosystem with the industrial Internet of Things (IIoT). Among them, healthcare is also broadly unified with the Internet of Things to develop an industrial forthcoming system. Utilizing this type of system can be facilitating optimum patient monitoring, competent diagnosis, intensive care, and including the appropriate operation against the existing critical diseases. Due to enormous data theft or privacy leakage, security, and privacy towards patient-based informative data, the preservation of personal patients’ informative data has now become a necessity in the digitized community. The produced article is underlined on handsomely monitoring, perceptively extracted keyframe, and further processed lightweight cosine functions using hybrid way chaotic map keyframe image encryption. Initially, a regular concept of extracted keyframe is deployed to salvage meaningful detected frames by transmitting an alert autonomously to the administration. Then, lightweight cosine function for encryption is employed. This encryption incorporates keyframe exceedingly secure and safe from the outside world or any adversary. Our proposed methodology validates effectiveness throughout the IIoT ecosystem. The produced outcome is highly acceptable and has minimum execution time, robustness, and reasonably adopted cost-effective, secure parameter than any other (keyframes) image encryption methods. Furthermore, this methodology has optimally reduced bandwidth, essential communicating price, transmission cost, storage, and immediately judicious analysis of each occurred activity from the outside world or any adversary to remain secure and confident about the real patient-based data in the smartly developed environment.


Introduction
Nowadays, the Internet of ings (IoT) has intensified its global omnipresence. e implementation of smart networks is supposed as an exposition of ubiquitous computing. e goal is to expand network edge enabling smart services with IoT system. ese kinds of computing is the best way replacement of the upfront user intentness towards interconnected employed devices (sensor-enabled strategies). It deprived of human interaction and instigated industrial perspective such as industrial Internet of ings (IIoT) [1][2][3][4][5][6][7][8]. ese applications are well-organized integral sensor-based IoT architecture which is behaving well-informed accosted systems, such as smart cities (SC), smart healthcare system (SHM), smart wireless-multimedia sensor network (SWMSN), smart homes (SH), remote monitoring on farm animals (RMFA), automobiles, drone monitoring system (DMS), smart industry surveillance system (SIS), agricultural crop monitoring system (ACMS), pain monitoring system (PMS), self-driving vehicles (SV), and smart transportation system (STS) [9][10][11][12][13][14][15][16][17]. ese treasured research and innovation parturitions as a high processing skill set in terms of the intelligent operational IIoT ecology, which is well suited to meaningful communication through the environs.
WMSN (wireless-multimedia surveillance networks) are among the top innovative contributor. It is an essential parameter towards IIoT empowered ecology that operates visual sensor data uniformly. It is also measuring all the possible views and nonstop capturing visual images. ese laboring methodologies are generating enormous multimedia visuals with too much redundancy into the system [18][19][20][21][22]. Due to availability of huge redundant visuals into the intelligent system, it is observed consensus of the researchers that developed a mechanism which can process meaningful as well as informative visuals from the surveillance networks. Add the best way recorded for forthcoming usages. It is tracing behavior or activities (simple interaction or abnormal), diagnosis liveliness of the doctors, operation theater activities, patients handling ventures of the staff as well as nurses, and observing hygienic facilities in the whole industrial healthcare setup. e approaching mentioned problems. e optimum emphasis is to capture every probable abnormal activity detection by accurate analysis with well-organized supervision and video abstraction. e foremost motive behind this, transferring all the visuals through the communication medium before processing is not the best way because it contains more bandwidth as well as energies. Additionally, it is very hard (difficult) along with time-consuming to recognize and ingeniously extracted action-oriented intelligence from a high volume of surveillance visuals [23,24]. erefore, it is essential to employ a piece of machinery or methodology that can accumulate each valued visual information individually incorporating high processing skills set and communication capabilities of the smart-enabled IIoT sensors. e quality of these methodologies is intelligently selecting accurate views from any different locations or multiple views and smoothly captured. e closely informative pursuits or core-specific data in real-time with the sensors, as a result, accurately transfer that visual data to the expert witness. Additionally, the key importance is to take action from the specialist by the investigation of original gigantic enlightening visuals. en, a conventional methodology including efficient monitoring (surveillance) is shown in Figure 1. e enhanced conceptual framework is apprehending each perilous movement for accurate recognition of any possible happened actions in a quite shrewdly and reporting immediate to enforce action promptly and reduces any miss happening into the entire system. is approach also provides healthier achievement regarding disciplined resource utilization and robustness, which is the essential requirement when monitoring smart healthcare systems with the (WMSNs) wireless multimedia sensor networks by proper investigation to resolve any technological, nature-based, and human malfunctioning defects [25,26].
Concerning the tendency of visual transmission at wireless environments as a WMSN from start to endpoint, vice versa, these communications are coming in a vulnerable way with the enormous security threads. Consequently, it is highly recommended that the visual data in terms of keyframe image can be securely transferred to a base station with enhanced and secure guided way without any modification from any untrusted parties. Additionally, we are guessing that the utilized dedicated communication of visual data supposed to problems due to jammed (congested) spectrum allocation methods into WMSN medium. Furthermore, the unfolded report is an emphasis on solving mentioned problems by incorporating an intelligent and power-efficient system. at can easily handle in better way by gathering informative data or relevant information in real-time and prompt action taken against the happened events. By employing this, our methodology can be proficient in reducing the cost of the transmission as well as consumption of congested bandwidths. In addition, adapted methodology has the primary intent to enrich the security prototype, in which system is fully taking care, protecting WMSNs, and improving utilization towards concern authority. Technically proposed system engendered encrypted keyframe images when transferring extracted keyframes to concern authority for accomplishing high rate of security during entire communication inside the smart healthcare IIoT enabled setup.
Core enrichment of the designed report is registered as follows: (I) We proposed an effective architectural adjustment to monitor healthcare smartly with IIoT via utilizing cosine as a function targeted encryption based on chaotic hybrid map. (II) e methodology emphasizes first to extract the most relevant or meaningful keyframes from the summarized video data into the keyframe extraction model. 2 Scientific Programming (III) After that, we incorporated a well-organized probabilistic as well as lightweight cosine function for encryption. is encryption approach is conquering strong security against any adversary. (IV) e methodology is applied to hybrid technologies such as Python, TensorFlow, YOLOv3 for the extraction of keyframes, and cryptographic simulation done by MATLAB. e extracted keyframe evaluation and security analysis ensured that our methodology is proficient for reducing the cost of transmission as well as the efficiency of consumption bandwidth. (V) e produced report endorses the commanding characteristics of the patient-based privacy in terms of an encrypted matrix to avoid any adversary outbreak. (VI) e produced report also approved that the numerous rigorous security threads can withstand.
e rest of the article is summarized and pursues as follows: Section 2 has articulated preliminary work briefly about the smart healthcare system, monitoring and surveillance system, video summarization, and RGB image encryption. Section 3 has incorporated a novel planned monitoring (surveillance) tactic in which model extraction of keyframe from visual sensors is briefly discussed, and it describes implementation of lightweight cosine function hybrid chaotic map encryption methodology. Section 4 has investigated its experimental outcomes and relative discussion. Section 5 has briefly encountered numerous rigorous security examination with each possible securitybased parameter. Section 6 is an efficiently concluding move toward entire work as well as a future direction.

Preliminary Work
Recent huge progressions in the territory of WMSNs are owned as a smart intelligence-based healthcare setup. It elevates finest incorporation at the hospital management with relating to patient privacy, security, and safety in the industrial perspective. Targeted mensuration about confidentiality, safety, and security, it is important to examine visual intelligence data and maternal encryption methodology.
at can validate the genuineness of the system designed for complete patient protection or towards health safety. Although providing relief to human lives, the huge surveillance information poses the challenging task of spending time and energy while the footage is being picked up. Consequently, some methodology or approaches are required to facilitate as well as provide the most relevant data as a summaries video, including extraction of meaningful keyframe within that summaries video instead of viewing the whole surveillance visual data. e video summary received considerable study coverage over the past two decades. In a very concise way, it helps to abbreviate one long video or different images, such as video skimming as well as a static storyboard. Khan et al. reported on [27] an effective, coevolutionary neural network-based summary procedure for Scientific Programming 3 a resource-constrained system of the surveillance videos, in which the shot segmentation process is incorporated by utilizing deep features. at mechanism retains the interesting nature of the produced summary by utilizing image memorability as well as entropy properties. It also claimed that in each shot of the frame recorded highest score as a keyframe of the memorability as well as entropy. Hussain et al. [28] expressed his thought about MVS (multi-view video summarization), how it is challenging to accommodate gigantic volumes of data, light variation, redundancy, overlapping views as well as inter-view correlation as well. eir idea is to integrate MVS with a deep neural network as a two-tier method to adopt soft computing processes. Its first automated tier conducts segmentation of priority shots based on appearance and preserves them in a query table which is forwarded to all the cloud with further analysis. Its second tier captures deep properties from every frame of a series in the indexed table as well as transfers those to deep, long-term bidirectional memory to gain insightful odds and providing a summary.
Huang and Wang [29] explained the popularity of video summarization and its key point selection of the keyframes that can represent actual content as a video sequence. eir idea is to provide video summarization as well as motion of the video summarization. For achieving this, initially the capsule networks are accurately trained, such as extractor of spatiotemporal details, as well as focused on such spatiotemporal properties as an interframe movement curve is produced. A (TED) transition effect detection system is subsequently proposed to segment the video streams automatically into shots. At last, a self-attention framework is implemented for selecting keyframe sequences within videos. erefore, selecting key static images as a summary of streaming content as well as measuring visual flows as a summary of video motion. Jei et al. [30] proposed an actiondriven video synthesis paradigm focused on reinforcing learning. e framework is divided mainly into two sections: video cut through action parsing and video description focused on reinforcement learning. Within the first section, a chronological multi-instance learning framework is equipped using weekly interpreted data to crack the issue of the time-consuming maximum annotation as well as the uncertainty of the weak annotation. roughout the second section, it built a deep recurrent video summarization model constructed on the neural network that selects the most recognizable frames compared with other behavior. In the meantime, the consistency of the mainframes extracted may be assessed by the correctness of the categorization. Yuan et al. [31] introduced a new (DSSE) deep side semantic embedding prototype to produce video summaries using the free side details. e DSSE establishes an implicit subspace through correlating the two unimodal autoencoders' hidden layers. Respectively, that embedded the side information and video frames. Furthermore, by dynamically reducing the loss of semantic meaning and the loss of the two unimodal autoencoders in function reconstruction, the comparable general facts between side information and video frames can be more thoroughly understood. Hence, its semantic significance can be evaluated quite efficiently. Ultimately, semantically relevant portions are picked from videos by eliminating its lengths to the side details and the latent subspace is constructed. e cryptographic encryption of the digital images can be extensively recognized into two major groups. One of them is color image encryption, and another one is grayscale image encryption. e encryption methodology of a grayscale image is only based on one plane. But it can be prolonged by incorporating different color planes like red, green, and blue to adopt completely color image encryption procedures. Digital security is a vital problem limiting IIoT applications as well as cyber-physical systems' wider acceptance. In these regards, focusing various cryptographic concepts earlier, Wang et al. [32] expressed their methods to incorporate quantized logistic map-based stream encryption among the (WBAN) wireless body area networks. WBANs are committed to the transmission and processing of human-grown biomedical information.
is encryption mechanism uses a chaotic method to execute the process for encryption.
e assessment results demonstrate that the suggested encryption framework seems to have the advantages of success with high-security defense. Al-Khedhairi et al. [33] reported their hybrid cryptosystem for achieving three goals. One of which is to implement a new (2D) twodimensional fractional-order map with very dynamic chaotic behavior as well as a unique true value of Lyapunov exponents across a wide variety of variables, particularly in comparison to other 2D maps. Secondly, for the first time, it is suggested a new robust, stable encryption system integrating the related chaotic pseudo-orbits of the suggested map with those of the benefits of elliptic curves through public-key cryptography while implemented to color images. At lastly, the hybrid structure is confident of verifying secure exchange of hidden keys as well as extremely obscure and concealing information communications. Zhang et al. [34] described an effective S-box, secure hash algorithm MD2, and fractional-order logistic map-based image encryption. Furthermore, the authors emphasize that the proposed algorithm is focused on a logistic fractional-order map which had benefits in their improved potential to withstand specific cryptanalyst attacks with reduced execution times. Huang et al. [35] articulated about the procedure for the color-based image encryption wherein permutation-diffusion happened concurrently. In addition, it is cooperating strangely among the color image matrixes. Accordingly, this approach is added security characteristics through high competence.
Almalkawi et al. [36] incorporated a lightweight image encryption method in which a hybrid kind of chaotic maps (logistic and henon) are used to encrypt the image matrix. His idea is to split digital images into blocks and effectively compress those blocks for reducing the size of the image matrix. After that then applied a logistic map for the creation of the key. e permutation and substitution operations are performed to attain shuffling and transportation of the image pixels. e henon chaotic map is implemented to adjust the pixel positions throughout the diffusion process to improve the necessary level of protection and to resist different security threats. Chai et al. [37] explained an RGB image crypto-method founded on chaos as well as DNA oriented encryption. Firstly, his idea is to crumble color image as a red, green, blue channel. After that, intra-inter component permutation methods are reliant on a plain image which is functioned to scuffle them. Additionally, transformed and rejoined permuted classification in DNA matrix. Lastly, ornamental security employed a twice confusion process through scrambling images in receipt of encrypted images. ey were creating pseudorandom chaotic orders, and the author accumulated a four-wing hyperchaotic process in the methodology. Wang and Li [38] proposed an innovative amalgamated chaotic color image encryption by using logistic and tent maps for preliminary key scanning. is method utilizes a chaotic hybrid map together via a structured logistic map and tent map. Afterward, it receives the necessary specification for Arnold mapping via a functionality transformation required to scramble that image matrix. e diffusion process utilized a chaotic neural network via Hopfield to produce a chaotic sequence of self-diffusion. At that time, the key is also created by a parameter reconstruction. Eventually, the mashed image is operated XOR with a key to acquire the penultimate encrypted cipher image. Hamza et al. [39] expressed an effective predictive cryptosystem to preserve the confidentiality of keyframes and the privacy of the patients. An innovative PRNG (pseudorandom number sequence) focused on chaos, concentrating on merging as well as cascading the positions of two of those same chaotic maps. By attaining safe and privacy-conserving communication, the medical professionals diagnostically enforced keyframes through limited bandwidth of energies and communication. Investigational testing from various viewpoints is done to ensure a high level of protection with more effective implementation relative to current methods. Hua et al. [40] proposed a CTBC (cosine-transform-based chaotic) system, in which two chaotic seed maps are used as a chaotic hybrid map to produce dynamic and a complex pattern of sequences. e encryption mechanism takes high-efficiency scrambling to isolate neighboring pixels and operates arbitrary pattern substitution to disperse a very small shift in the images towards all pixels of cryptographic matrixes. Additionally, the summative assessment shows that although the chaotic maps created by the CTBC demonstrate far more complex chaotic responses than the currently available.
e experiment result shows how reliable the suggested image encryption framework is.
Hamza et al. [41] incorporated a safe system for video summarization of exterior patients under WCE protocol. In this method, keyframes are derived using only a lightweighted video description framework to make it fully WCE-fit. After that, a cryptosystem focused on Zaslavsky chaotic map. It is provided for the protection of derived keyframes. Observational findings confirmed the suggested cryptosystem's efficiency with respect to reliability and highlevel protection relative to many other modern image cryptographic algorithms. During most of the distribution of essential keyframes to customized WCE healthcare providers and gastroenterologists, Hamza et al. [42] reported an effective cryptosystem for IoT-based monitoring systems.
Firstly, a lightweight automated summary methodology is utilized. To retrieve the keyframes throughout the surveillance footage, based on a simple histogram-clustering method, then, it compressed the synthesized data size using a discrete cosine transform (DCT) methodology. Eventually, the implemented architecture executes effective image encryption with the introduction of discrete random fractional transformation (DFRT). Kaur et al. [43] incorporated pseudodominated genetic algorithm processing and specific chaotic image encryption techniques. It is introduced to change the 5D chaotic map hyperparameters. Firstly, the incoming image is broken down into thread bands that used a dual-tree dynamic wavelet transform (DTCWT) to perform the TFCM. erefore, these thread bands are diffused, and the hidden key is used from just the engineered 5Dchaotic diagram. An inverse DTCWT is gradually enforced to get the ultimate encrypted image. Broumandnia [44] implemented 3D modular chaotic mapping that enhanced key size and acceleration for encryption of RGB images. To extend key space color picture encryption with modular arithmetic, he provided reasonable success with the study of the histograms as well as a correlation of the adjacency pixels.
is method is used the confusion and diffusion function with the image encryption measures implemented in substitution and permutation.
Mondal et al. [45] introduced a highly efficient image encryption system for safe interaction and preservation of images. e framework is concentrated on a map of the chaotic skewed tent as well as cellular automatons (CA). e synthesis for both chaotic maps with CA respectively provides a model which a larger key size as well as quicker generator PRNS. e results of the experiment indicate good effects of encryption that can also withstand every kind of identified attack. e proposed concepts followed an additional parametric observation presented with the distinguished modification, which reflects a relatively faster encryption methodology at [46].

Proposed Work
e mounting standard of autonomous monitoring is highly progressed manufacturing-based visual sensor and technological progression of IIoT. It is bringing effective management tools that can make accurate as well as easy analysis of the digital world. It can also perform constant adoption of cumulative monitoring networks. is can be well-known as a smart setup of healthcare. ese technological ecologies are empowered to be analyzed autonomously in an intelligent way in real-time of visual data (video source). VSN (visual sensor networks) have the capability to interact smartly and agreeing to conduct critical as well as very complex visual data processing in real-time through the processing capacities and increased storage. For multi-view observation videos recorded in the healthcare communities. Its computing competencies can be utilized to evaluate streaming video footage to categorize keyframes as well as afterward eliminate obsolete and meaningless visual data, thereby reducing the parameters for minimizing bandwidth. e advanced communication skills of sensor nodes can also be Scientific Programming utilized to cooperatively accomplish comprehensive scene assessment to produce real-time multi-view overviews of monitoring keyframes.
Highly intelligence-based sensors can be utilized to yield a mechanized response. Subsequently, anomalous events are spotted such as the need of the patients in emergencies, discomforting from high and low blood pressure among the patients, and feeling symptoms like heart attack among the cardio patients, unbearable pain among the patients, and required help by shouting, monitoring high-risk patients' activities like special care of cancerous patients treating in the special wards [47]. Besides, the core emphasis of our methodology, it is incorporating a very well-designed lightweight keyframe image encryption among the extracted keyframe. A well-organized, highly acceptable proposed smart setup healthcare IIoT is presented in Figure 1. Further coming sections are enforcing the best guidelines for achieving required results and its essential personifications.

Concept of Extracted Keyframes from Visual Information.
In the concept of keyframe extraction, visual information is transversely produced by installing visual sensors and retrieved from VPH (visual-processing hub). is is retrieved in terms of video frames into the entire smart setup which consists a massive amount of visual data. ese visual data transportations are unrealistic in terms of distance covering (more data traveling) in entire networks due to bandwidth and limitations of energy constraints in the wireless multimedia sensor networks. To handle these problems, many researchers have employed some technique to control and limits such a massive video information flow. Regarding this, researchers are adopted distinct compression schemes [48,49] as well as video summary approaches to minimize a huge amount of visual contents at the visual-processing hub. So as to the appropriate frames, videos are traveled towards base stations [23]. Considering energies and bandwidth constraints, we employed an energy responsive keyframe extraction methodology to diminish data redundancy [50]. Our extraction methodology consists a lightweight keyframe extraction with an improved YOLOv3 algorithm [51]. In this methodology, our main emphasis is on extracting relevant keyframe from the summarized video, which is coming through installed wireless multimedia surveillance networks in the smart healthcare system. is algorithm adopted a backbone network which is called Darknet-53 [52].
In Figure 2, the up-sampling network, as well as detection layer, is recognized as a YOLO layer. YOLOv3 is completely dependent on Darknet-53 to extract keyframes from the visual input data. As an essential component, each network is using a residual. Each five layers of residual are selected differently in terms of depth and scale to perform residual among the produced output at diverse layer. All the convolutional in the number showed 53 with a residual block in terms of pair 3 × 3 and 1 × 1 layer. e spatial resolution is 1/32 smaller compare to visual input data at the final feature map size. YOLOv3 recognizes as a three-scale layer of YOLO which is basically answerable for the detection of the object at a different layer of scale. e First YOLO layer is based on grid resolution. It contains 1/32 of the visual input data and detecting big objects. e last layer has 1/8 resolution, and that can effectively detect a small extent of objects. Between the layers, one known as the upsampling layer, and the other are several convolution layers that exist [53]. e resolution of this algorithm is initially set by default 416 × 416, and the algorithm is also favored height, as well as width, must be equal in size. But no essential to modify resolution before inputting images. We had designed appropriately in a manner to adjust the resolution of each image automatically when its obligatory. e proposed extraction approach is administering a conceptual description at Figure 2.
e whole setup includes a combination of four training modules. Firstly, the interaction of visual contents towards training module is defined. is module has been defined to train data with the help of the model module. Secondly, model module is responsible for modeling data towards received from first module.
is module has capability to further model if there are some deficiencies from earlier process. irdly, modeled data are approaching to prediction module for accurate prediction. At the last foresight, prediction module is transferred data to detection module. Finally, the output comes as a human presence (patients and staff) via recorded video summary in terms of keyframe. A lightweight YOLOv3 algorithm [51] is quite efficient to detect visual data, either videos or images, in a real-time manner. For the proposed extraction of keyframe, we properly trained the model in the Darknet-53 platform [52]. Additionally, this model is also renewed into the TensorFlow atmosphere. Attaining high precision, which is much required for the experimental work, we trained this model with a vast collection of image datasets. One of them is a wider-face dataset [54]. Originally this model is initially set floating-point as a by default. erefore, it is transformed in the fixed-point model through TensorFlow. By converting the model, this model behaves computationally competent approach. It is effortlessly incorporated into the smart healthcare system. is can also suit any such small devices such as visual sensors that consume processing, bandwidth constraints, and energy. We effectively tested the model of the extraction by utilizing Face Database (FDB) [55].
Our incorporated methods are proposed to increase gratitude precision at encouraging real-time exhibiting YOLOv3s bounding box, which is a more appropriate symbol of SSD (single-stage detection). e experimental results of the average accuracy are received 88-90% with 1-16 files per second (FPS). is model is operated at the environment of Intel (R) Core i5-6500 Microprocessor, CPU (Central Processing Unit) 3.20 GHz, RAM 8 GB, functioned Windows 10 Operating System. at is highly suitable concerning patient-based monitoring into smart setup healthcare IIoT-enabled architecture which is revealed extracted keyframe in Figure 2. is also indicates that the patient is correctly detected in terms of high precision as well as inside the bounding boxes. e process of extraction is incorporated among the patients for testing numerous clinical wards with the proper settings. After significantly produced keyframe from the model of extraction, the 6 Scientific Programming keyframe is passed to lightweight cosine function through hybrid chaotic map encryption for the additional security process.

Lightweight Cosine Functions with Operation of Hybrid Chaotic Map Keyframe Encryptions.
is subdivision acquaints with examines the planned cosine function of chaotic sequence (CCS) [40]. It is physiognomies along with effective encryption methodology into the smart healthcare IIoT ecosystem. It is caused by keyframes from the video streaming data by the visual sensor. Our approach is handling first to generate PRNG from CCS with the incorporation of a secret key into the keyframe images. Secondly, it performed confusion as well as diffusion operation into the keyframe image as an engagement of bitwise XOR and highly efficient scrambling algorithms [40] as shown in Figure 3. is methodology is achieved by fast RGB keyframe encryption, randomized complex sequences, and highly encrypted cipher keyframe data. is cipher matrix is impossible to recognize an adversary to detect actual keyframe images. is can also confine the approachability of the information required in a typical cryptosystem. ese keyframes are emanating from the monitoring of video streaming, which is done in terms of RGB format with highly resolution visual sensors installed at smart setup of IIoT enabled healthcare.

Preliminaries about Cosine Function.
e CCS has incipiently pleaded each existing drawback that frail in chaos as well as tendencies of the weakness in the chaotic maps.
is can be explained better in terms of used mathematical notations as below: where Chaotic Seed Map 1 and Chaotic Seed Map 2 are known as two different used chaotic maps. at are also well known as a combination of seed map with some control parameter and constant shifting β (set as β � − 0.50).
Equation (1) enlightens that CCS integrates with two seed maps output with adding constant shifting operator to get cosine transformation in terms of output. Mixing with two seed maps are the outcomes of a complex and efficient sequence generation in the cosine function-based hybrid chaotic sequence (CCS). Using two seed maps can also provide flexibility in terms of making countless fresh chaotic maps employing different configurations of prevailing maps.
ere are so many chaotic seed maps available that can be a better alternative as a complex sequence provider with the consideration or nature of the requirement. Some of them are mentioned below throughout: Logistic Map: Arnold Map: Tent Map: Chirikov Taylor Map: Sine Map: Zaslavskii Map: Henon Map: Arranging the above chaotic seed maps in terms of haybrid nature such as Equation 1. e incorporated cosine function-based hybrid chaotic sequences compile on the following: Cosine-transform: Pair 1. Scientific Programming Cosine-transform: Pair 2.
where r ∈ [0, 1], α replaces r and b swaps 1 − r. Equations (9)-(11) are the same as our standard notations, as mentioned in equation (1). e proposed method is used equation (10) STC maps (Sine Tent Cosine) as the actual combination of the CCS in the entire paper, which is the adaptation of Sine and Tent seed maps.

Structural Setup towards Producing Key.
e reliable key controls the series of STC at primary states. Concerning the ideal state of key, the researchers emphasized on [56]. It is highly counseled to withstand diverse sorts of attacks whenever the key size of every chaos-based cryptographic protocol corresponds to the 2 100 proportions. Concerning the effective length, which can provide freedom to maintain a higher security aspect, the key space of STC-IES used 256-bit long proportion that can be represented as 2 256 . It is made of five key components such as K � {X 0 , Y 0 , U, g, d}, in which preliminary states are addressed with (X 0 , Y 0 ), U is the best represented component as a constraint of disorder that befuddle the preliminary states, g is the coefficient of preliminary states, and d comprises four-parameter of disturbance coefficients represented as {d 1  ese preliminary stages can be additionally calculated by the way of proclaimed setup of encryption operation form as follows: e STC map is generated uniformly scattered chaotic sequences, which is shown in Figure 4 through Table 1 for the process of bitwise XOR and highly efficient scrambling at the preliminary stages (X 1 , Y 1 ).

Steps of Lightweight STC-IES.
e proposed STC-IES approach is the keyframe encryption as well as decryption algorithm suggested in Tables 1-3, respectively. e key approach for producing encryption is demarcated as follows: (I) Each generated keyframe image is first permuted by sine tent cosine (STC) chaotic sequence through a randomly generated secret key to accomplishing a more complex chaotic sequence. Additionally, keyframe is reshaped through recently created complex nature chaotic sequence. After that, splitting each keyframe image as a matrix form of its color RGB channel such as red, green, and blue, an operation is performed as a bitwise XOR and highly efficient scrambling. (II) Each matrix of color keyframe is diffused by bitwise XOR by combining highly efficient scrambling algorithms to generate the cipher keyframe images. (III) e cipher keyframe is confused through the highly efficient scrambling procedures [40] to obtain the caused encrypted keyframe image.
e decryption process approach is fully followed through the inverse of highly efficient scrambling procedures and inverse bitwise XOR methods, that is, consecutively decrypt encrypted cipher image at the original keyframe images. Tables 1-3 are confirmed pseudocode. It is used for bitwise XOR, STC sequence generation with the addition of highly efficient scrambling algorithms, one-toone, where S is the sine tent cosine (STC) chaotic sequence and the size of the keyframe image is M × N. e produced secure key is used in entire cryptosystem and is formed with the random key generation procedures.
10) end of the for loop (11) STC generation Table 2: Algorithm 1, encryption process through bitwise XOR. Input: size of keyframe image (M×N) per initial state (X i 0 , Y i 0 ) Output: encrypted cipher keyframe C (1) Read size (M×N) RGB keyframe image (2) Resizing keyframe images (3) Generation of random number for getting key (4) Generating secret key by calling function (5) Generating STC sequence by calling function (6) Reshaping keyframe images with STC (7) Splitting RGB keyframe image in each 3 channels IR, IG, and IB (8) Bitwise XOR operation is performed in every channel (9) Performing highly efficient scrambling algorithm obtained keyframe channel from step 9 (10) Merging every scrambled keyframe channel (11) Encrypting cipher keyframe images C

Simulation Results and Discussion
is segment imposes as well as discourses the investigational consequences of the proposed STC-IES on MATLAB 2018a version software. It also analyzes security with the help of using test images from the eminent source USC-SIPI [57] repository database of digital images. An ideal and prominent encryption approach is always intelligent to encrypt dissimilar classes of images into highly secure cipher images. When an encrypted method is strongly enforced to reach an optimum security level in terms of cipher images, it is noticeable that any keyframe can only be retrieved by knowing the arcuate used top-secret key. Unless the known used secret key, not any information of the keyframe images can be extracted in smart healthcare IIoTenabled system. Figure 5 is illustrated the encryption approach. It used USC-SIPI images A (as a test image). Image A is the pepper image, and its corresponding originated all the threecolor frequency as mentioned at histogram R, G, and B, which is purely showing the maximum correlation of the pixels in each color channel. e encrypted cipher image of the pepper image B and its corresponding three-color channel histogram R, G, and B show uniform distribution in each color channel. Similarly, Image C is the keyframe image. It originated from all the color channels as shown in each color channel histogram R, G, and B, which originally showed maximum pixels of the correlation in individual color channels. e encrypted cipher image of the keyframe image is D. It is conforming that the entire three-color channel histogram R, G, and B are screening uniform distribution in the separate color channel. Figure 5 is indicating that the proposed approach has a strongly encrypted methodology. Uniformly distributed pixels are also visible in the cipher image. It means this cipher image has adopted better encryption process; in this way, any intruder or attackers cannot effectively gain original keyframe information from cipher image. erefore, STC-IES can easily convalesce every keyframe fully from consequently encrypted image data for the reason that all the handling process is completely reversible. Our methods have robust inefficiency, fast execution, and minimal computational overhead, employing bitwise XOR and highly efficient scrambling algorithms at a smart healthcare setup. us, this encryption methodology can be accomplished faster and efficiently in the IIoT ecosystem. To provide better security as well as privacy towards the patients in the respectively possible emergency requirement, Table 4 depicts the time complexity to encrypt keyframe, uses test images, and undergoes comprehensive comparison with the different types of available encryption algorithms. e simulation and experimental cryptosystems are effectively conducted on the required platforms. For example, Intel (R) Core i5-6500 Microprocessor, CPU (Central Processing Unit) 3.20 GHz, RAM 8GB, operated Windows 10 Operating System. e demonstration of the fast keyframe encryption is exhibiting STC-IES from various image extent exemplified on the numerous results in terms of tabular form. It is always required to have a vintage highquality of deciphered images because a noteworthy discrepancy of pixels into keyframe can disturb entire pixels into the encrypted images. Incorporation of the STC-IES can be a safeguard in terms of higher security aspects of the cipher images. A delicate modification of pixels into a ciphered encrypted image matrix can distress some pixels as a deciphered conclusion (results) into the decryption approach. In this crucial condition, if STC-IES encrypted data miss some pieces of information, the process of decryption can also easily recuperate tangible keyframe visual facts. Figure 6 demonstrates the quality-based approach after the decipherment process as soon as the STC-IES encrypted cipher images suffered from noises or various types of proportion data forfeiture. For example, in Figure 6(a), the decipherment methodology is fully recovered in pepper image and no data loss into encrypted cipher of pepper image are viewed. Even though the cipher images had also lost some data (info) or noise moreover, their deciphered findings incorporate most sensory information of the same individual pepper images, which can be presented in Figures 6(b) to 6(d). Accordingly, the employed STC-IES can fairly decrypt the superior quality in terms of encrypted cipher.

Security Analysis
With the intention of validating STC-IES preeminence, this unit scrutinizes efficient security in terms of resulting aspects with minimal encryption speed assessment, histogram analysis, information entropy analysis, differential attack analysis, correlation analysis, analysis of produced key, and comprehensive comparative analysis among surveillance systems. To supplementary exemplify the effectiveness of STC-IES, we equate our approach with the  (14) m � Si, j; n � Tm, j; (15) c 1 � ((m − 1)÷H); c 2 � ((n − 1)÷H); (16) b 1 � (m − 1)mod H ; b 2 � ((n − 1)mod H)+1; (17) x � c1×H + c2 y � b1×H + b2 (18) Scramble channel (x, y) � operated keyframe of bitwise XOR channel (i, j) (19) end of the for loop (20) end of the for loop 10 Scientific Programming other available advanced image encryptions. We cited directly, respectively, finding in terms of references reported at esteemed scientific journals by the author for the fair comparison at the area of highly noticeable imagebased encryption algorithms.

Encryption Speed Assessment.
is segment is familiarized with minimal computational-based overhead and speed assessment displayed in Table 4. e time computation segment of any encrypted system depends on the generation of chaotic sequences with the proper handling of permutation and diffusion methods in the operating algorithms, eEnforcing smart healthcare enabled IIoT system. We accentuated to achieve minimal computational overhead for constructing relatively better communication in speed with the adoption of the STC-IES mechanism. We observed that the generation of complex sequences is relatively quick in nature. e operation of bitwise XOR with highly efficient scrambling is operating in minimal computational time. e proposed approach is demonstrated as an average encryption time assessment for each set of various sizes of the keyframes. e production of numerical values of the encrypted keyframes is presented in Table 4 and compared with other past color image encryption methodology such as [20,37,40,[58][59][60][61][62]. Our proposed approach  is fast in running and has minimum computational effort, which enforces for real-time application setup like a smart healthcare IIoT-enabled architecture.

Histogram Analysis.
Keyframe histogram is the best illustration of a typical graphical representation of the pixel rate distribution in the keyframe images. e keyframe/encrypted cipher keyframe image histogram contains the statistical data by which evaluation of the robustness recognizes the statistical data analysis. Really, histogram reports about the distribution of a keyframe's grey-level values, and relatively smooth distribution will expose the loopholes in the method that attackers can use to initiate a preferred-ciphertext attack using statistical analysis. However, for reliable cryptography, the dispersal of histograms must be uniform. Figure 5 demonstrates the histograms of the keyframe and its ciphered images, which can pictorially recognize that the arrangement amongst pixel correlation at ciphered-keyframe images has fairly unvarying color frequencies. Still, there are a few variations in the original keyframe image [44,58,63]. e histogram deviation is used to measure an encrypted image, and this also contributes to significant analysis. Suppose the quantity of variation in the encrypted cipher keyframe image is small, the greater its uniformity. Two encrypted keyframe images are produced using different surreptitious keys, even with similar image data. It shows better homogeneity, uniformity, of the encrypted keyframe images if the differences are similar enough. Figure 5 demonstrates color test image histograms, wherein it is clearly verified, specifically color frequency histogram representation and its encrypted image histogram accurately. We perceived that each color channel of the test images and keyframe images shows their natural behavior before encryption, and after encryption, each color channel behaves uniformly in nature. e uniform behavior of the histogram can be calculated numerically of a variance. e uniform behavior of a histogram is purely dependent on the variance. Lower variance is recognized as the higher nature of uniformity. e higher variance is lower uniformity. e variance of the histogram can be statistically obtained by Eq (13), in which grayscale numerical value is n, Y i , and Y j stands for the values of the pixels at the i th , j th gray levels [58,64]: Consequently, in such a scenario, if any attacker decrypts any piece of the image, assuredly, the attacker cannot find the entire original data because the proposed method has a powerful encryption methodology that keeps away any data leakage. In the same way, with the observation of the keyframe channels as well as encrypted keyframe channels remains identical keyframe, it is signifying no data loss during the communication into smart healthcare IIoT enabled system occurs. erefore, the histogram analysis equally justifies that the proposed STC-IES methodology strongly avoided any statistical or numerical attacks. It also provided a statement for the integrity, including consistency during the transmission at smart setup healthcare system.

Information Entropy Analysis.
e methodology of analyzing the amount of uncertainty and randomness behavior among the cipher images' correlated pixels is called information entropy. It can be well demarcated as  Scientific Programming where information entropy is H (n) and P (n i ) is the probability of the occurrences at each n i . Generally, the possible information entropy's numerical value recognizes as 8 for the ideal scenario at each random keyframes or image. In an ideal, discrete image, probabilities of any containing n i signs are identical. e probability among each n i symbol is, therefore, 1/256: After encryption, the encoded cipher image must continue to act as the best possible random image.
is is validated that encrypted keyframe information entropy should be around numerical value 8, significantly proved from equation (14). e following is also demonstrated: the information entropy around 8 is highly random in nature and very low information leakage is present in our proposed smart healthcare IIoT-enabled system. Table 5 demonstrates information entropy of color channels with its extracted as well as encrypted keyframes (1)(2)(3)(4)(5)(6). It is also relatively compared in Table 6 with earlier referenced encryption methodology such as [20,50,[65][66][67][68][69][70]. Tables 5 and 6 exemplified by the projected approach have shown healthier results in terms of information entropy for each color channel, which almost have numerical value 8 as an encrypted keyframe image. is indicates that the proposed STC-IES methods deliver quite a suitable notch of security, privacy, and randomness behavior in contradiction to the information entropy attack.
at is the furthermost requirement for the appropriate way of handling secure concern smart healthcare IIoT-enabled system.

Differential Attack Analysis.
e differential attack is a well-known attack into keyframe image encryption. is attack is used as an active security attack. It focuses on associating a sturdy relationship among keyframes per their ensuing encrypted images via perceiving transformations at a keyframe. It can annoy encrypted data. If it preserves diffusion in an ideal way, the keyframe encryption methodology expresses high usefulness in the sidestepping differential attack.
is diffusion process proves that the encrypted-cipher keyframe possibly will scatter a minor alteration throughout the keyframe image over the entire statistics or data. e proposed STC-IES mechanism is appropriately demonstrated as an innovative way diffusion process in Figure 5. e differential attacks can be best examined among encryption methodologies to withstand attacks with the two well-known metrics such as NPCR (number of pixel change rate) and UACI (uniform average change intensity) [74]. NPCR calculates the change percentage amongst the pixel positions within an encrypted cipher keyframe by adjusting one pixel's worth at keyframe or original image. UACI evaluates the average strength difference between individual keyframes besides their respective encrypted-cipher images. Suppose two encrypted-cipher keyframe images denoted as C 1 and C 2 which have a slight one-bit discrepancy in their corresponding keyframe; at that moment, their NPCR and UACI can be governed from equations (16)-(18) as follows: e magnitude of the keyframe matrix is denoted as M × N. Equation (17) is calculated G(i, j), which is pixel transformation between corresponding encrypted keyframe images. Table 7 pointedly illustrates the best possible NPCR and UACI produced values from keyframe in the proposed STC-IES algorithm, and reasonable comparison is shown in Table 8 through previously referenced encryption algorithms such as [20,50,[71][72][73]75]. e results indicate that the produced result of NPCR is very near to a hundred and produced the result of UACI approaching very near to onethird of the hundred. is is purely demonstrates encryption methodology of STC-IES. It incorporates effectively dissimilar randomize keyframe images as well as completely misses the effectiveness of differential attacks. In this manner, the planned cryptosystem ensured complete security and privacy from the attacker for not getting any information into smart healthcare IIoT-enabled system.
For example, investigation of ideal encryption and randomness nature of NPCR, UACI result into cryptosystem. Suppose the encrypted matrixes which size is M×N, C1, and C2 are encrypted cipher keyframes. An ideal encrypted cipher image is a certain image that cannot be discerned from a pseudorandom image. Simply cipher image is an arbitrary arena at the size of M × N, where integer i ∈ [1, M] and j ∈ [1, N], arbitrary pixels price C (i, j) identical nature and independently occurs isolated unvarying distribution taking place 0 to C's principal buttressed numeral E such as ∀j ∈ [1, N] , ∀i ∈ [1, M], ∃C(i, j) ∼ i.i.d U(0, E) [74]. e Scientific Programming    14 Scientific Programming hypothesis assessment of NPCR (C 1 , C 2 ) by means of β-level consequence follows as [74] in It is implicit at what time NPCR(C 1 , C 2 ) < ξ NPCR [76], and H 0 hypothesis assessment is purely rejected. at is the vital worth designed for challenging NPCR. On the contrary, H 0 hypothesis assessment is acknowledged. e produced result of NPCR * β is critically enlightened [74] in the following equation: [77][78][79] is well known as CD (cumulative density) function of SD (standard normal distribution) which has range amongst zero towards one, expressed through N [0, 1], and E represented as a gray image that is incorporated numerical value 255 in the current study: Once UACI(C 1 , C 2 ) ∉ (UACI * − β , UACI * + β ) [80], H 0 hypothesis assessment is virtuously rejected.
at is the crucial worth designed for challenging NPCR. On the contrary, H 0 hypothesis assessment is legalized. e produced results of UACI * + β , UACI * − β are critically explicated [74] in the following: Tables 9 and 10, correspondingly, exhibit assessment results from level β � 0.05 of the NPCR and UACI.

Correlation
Analysis. CC xy correlation coefficient of the two neighboring pixels offered randomness information.
at is a parameter for calculating a keyframe cipher's robustness and can only be quantified utilizing equations (26)- (32). CC xy analyses the total amount of linear correlation between both the adjoining pixels in the keyframe images. For authentic images, each direction of the keyframe images (diagonal, horizontal, and vertical) amongst pixel and the corresponding pixel is highly correlated. e STC-IES method's primary function is likely to tinkle causal relationships between nearby pixels together at each direction, such as vertical (V), horizontal (H), and its diagonal (D). At the same time, keyframe data matrix achieved approximately zero correlation with the highest possible unpredictable nature and strangeness (randomness) [44,81,82]: e scope for the coefficient's performance is between +1 (a positive one) and − 1 (a negative one), minor the value of the coefficient for encrypted cipher keyframe image, higher the quality of the cipher to battle any the statistical attack. Scenario 1: If CC xy > 0, it exposes a positive correlation into the encrypted matrix. Scenario 2: If CC xy < 0, it exposes negative correlation into the encrypted cipher matrix. Scenario 3: If CC xy � 0, it exposes no correlation into the encrypted keyframe cipher matrix,

16
Scientific Programming [44,58,83]. In this article, Table 11 demonstrates the random assortment of ten thousand combinations of two corresponding pixels at every three directions and correlation coefficient (CC xy ) values for test image pepper, lake, and extracted keyframe images with the size of (512 × 512). e discoveries of Table 11 and Figures 7-9 state significantly that the CC xy of two adjoining pixels of extracted keyframe images in each direction, such as diagonal, vertical, and horizontal is just about to 1. However, the encrypted-cipher keyframe is almost zero. Produced result expressions of Table 11 including visual histogram representation at Figures 7-9, effectively validate the uppermost superiority of breaching the correlation link amongst the adjoining pixel at the test images including extracted keyframe images through planned STC-IES methods and fairly comparison on Table 12 through former referenced encryption approaches such as [39,50,66,75,84,85,87]. e results and comparison prove that the proposed STC-IES methodology is exceedingly unaffected to any sorts of statistical attacks into smart healthcare IIoT-enabled system.

Analysis of Produced Key.
Configuration among the chaotic seed maps is susceptible towards preliminary surroundings. A cryptographic scheme is classified of superior quality or appropriate key size if it has sufficiently computational complexity with such a heightened sensitivity to modify the key. e proposed STC-IES has 2 256 key size means 256-bit long proportion. at encounters the key performance requirements and is exceedingly operative in sidestepping dissimilar sort of security attacks [40,56,88]. Additionally, the proposed methodology, including key structure, behaves extremely sensitive to design architecture in five blocks. e STC-IES methodology articulated key size in Table 13 and equitably comparison with the past referenced encryption algorithms such as [37,65,71,[89][90][91][92][93]. It endorses that the proposed key size delivers a reasonably healthier variety of the key space to engender more complex chaotic actions. Consequently, STC-IES has the satisfactory key size to evade all the possibility of brute force attacks.

Comparative Analysis among the Monitoring and Surveillance System.
is subdivision is comparatively based on the proposed methodology with the heretofore referenced surveillance as well as image encryption algorithms in Table 14. Table 14 expressively illustrates all the key characteristics in terms of security surveillance and encryption methodology that can prove secure and robustness constraint based on key analysis, encryption speeds, entropy analysis, correlation coefficients (CC xy ), NPCR and UACI produced outcomes. e outcomes of the proposed methodology are fairly reasonable as well as nearly ideal standards. We compared proposed methods with past referenced schemes such as [20,39,41,42,50,71]. We noticed that the referenced surveillance and encryption scheme is produced relative results to attain confidentiality of the security. We compared our results with numerous images employing diverse platforms and upbringing functionality, each restrained factor. e proposed methodology has comparatively better speed with minimal complexity in the execution, analogous better-quality entropy, lowermost correlation coefficient (CC xy ), satisfactory NPCR, and UACI statistics. Significantly improved results are resolutely signposted. at the proposed methodology has exceedingly adequate in the arena of smart healthcare IIoT enabled setup with monitoring (surveillance) as well as cryptographically secure ecosystem.

Conclusion
e technological progression of the hybrid IoT ecosystem is referred to as the IIoTsystem. It is extensively organized with the industrial, medical healthcare system to deliver the finest services counting security and privacy about patients. e produced report formally focuses on security, privacy, and its challenges in smart setup healthcare IIoT architecture. We implemented secure monitoring (surveillance) with intelligently keyframe extraction and lightweight cosine function hybrid chaotic map encryption. It also assists, including security and privacy from the outside world or any adversary. At first, a well-disciplined model of keyframe extraction is employed with a lightweight YOLOv3 algorithm. is model is successfully operated and tested with a vast collection of image dataset Face Database to retrieve meaningful detected frames (normal/abnormal events) through the visual sensor. e average accuracy is received optimally around 90% with an acceptable frame per second rate. Second, a lightweight cosine function encryption is employed over approved extracted keyframe to remain exceptionally secure and safe from further attacks. e generated cosine function chaotic sequences are a non-linear transform to engender as well as exhibit expressively very complex chaos behavior. Our encryption methodology implies a better diffusion-confusion process of bitwise XOR operation and highly efficient scrambling algorithms, which are achieved to encrypt each color image channel separately and scatter neighboring pixels into different positions quickly. is proposed methodology ensured satisfactorily encrypted matrix as a cipher keyframe without identifying real keyframes into smart setup healthcare from adversarial threads.
Numerous security discussions and analysis results inveterate about the effectiveness of the proposed methodology. is has relatively higher security characteristics, and minimal computational processing agreed and contending with the earliest image encryption approaches. It also validates its accomplishment in the IIoT ecosystem with minimalizing bandwidth, storage space, communication cost, transmission expenses, and correspondingly waning time spending of specialists handling due to huge amount of monitoring data to take verdict over any such suspicious incidents detection or any suspicious action detection as an emergency need from the patients at smart setup healthcare architecture. e produced concept can also be utilized in many relatively analogous urgent responded real-time projects: traffic-control, fire-detection, crime-control, and smart transportation at smart cities.
For future concern, this approach can be supported to assimilate information from further systems, aimed at numerous applications and additional advanced security facet and privacy dealings in any exact areas within the healthcare sector. e innovative way can probably integrate dynamic key as an alternative applied process for additional ornamental security and privacy.

Data Availability
Previously reported are used to support this study and are available at [Wider-Face Data Set, Face Database (FDB), USC-SIPI Image Data Set Repository ]. ese prior studies (and datasets) are cited at relevant places within the text as references [51,52,54].

Conflicts of Interest
e authors declare that they have no conflicts of interest.