OBSERVATION SYSTEM RESOURCE PLANNING IN PRESENCE OF ACCESS CONTROL BASED ON VOLUME OF RESOURCE OCCUPIED
BY TRAFFIC FLOWS
DOI: 10.36724/2072-8735-2022-16-8-54-62
Umer Mukhtar Andrabi,
Moscow Institute of Physics and Technology (State University) Moscow, Russia, umer.andrabi@phystech.edu
The publication has been supported by the Russian Foundation for Basic Research, project No. 20-37-90048 postgraduates
Margarita G. Kanishcheva,
Moscow Technical University of Communication and Informatics Moscow, Russia, margo.kan@list.ru
Manuscript received 20 June 2022; Accepted 10 July 2022
Sergey N. Stepanov,
Moscow Technical University of Communication and Informatics Moscow, Russia, s.n.stepanov@mtuci.ru
Keywords: Radio Resource Management (RRM), Video Traffic Streams, IoT, Network Segmentation (NS), Surveillance System Operator, Access Control
Due to various recent improvements in cellular technology, communication networks now offer information flows with a wide range of transmission speed and service quality characteristics. These characteristics are more highly noticeable in intelligent networks, particularly when IoT technology is used. Regulation and high-quality transmission of such massive volumes of data present severe issues among virtual operators engaged in the communications sector and utilizing cellular network infrastructure to provide video surveillance services for its consumers. Unfortunately, no single approach exists that enables the execution of high-quality incoming traffic flow maintenance and the allocation of information transmission resources in diverse networks. One of the methods to address these issues is mathematical modelling, which takes into consideration the unique characteristics of traffic flows that emerge and are approved for servicing. Our work addresses this problem by designing a resource allocation strategy for an isolated LTE network cell with heterogeneous traffic flows while sharing its radio resources. Traffic sources considered are operator surveillance system's video cameras that use cellular network infrastructure to deliver data to analytical centers. A mathematical model is used to study the process of sharing a resource, which takes into consideration the priority of real-time traffic, the elasticity property during file transfer, and the access restriction for all threads, based on the quantity of resource held by each thread. When sharing the resources of an isolated LTE network cell with heterogeneous traffic flows. Traffic sources are surveillance system operators' video cameras that use cellular network infrastructure to deliver data to analysis centers. Algorithms for evaluating the characteristics of the service of applications are constructed and the dependence of the characteristics on the restriction of access is investigated. Using the developed model, a scenario of dynamic distribution of the access node resource between incoming sessions has been developed, which allows creating conditions for differentiated servicing of heterogeneous traffic flows based on access restrictions and resource loading by each of the considered flows.
Для цитирования:
Андраби УМ., Канищева М., Степанов С.Н. Планирование ресурса систем теленаблюдения при наличии контроля доступа, основанного на объеме ресурса занятого потоками трафика // T-Comm: Телекоммуникации и транспорт. 2022. Том 16. №8. С. 54-62.
For citation:
Andrabi U.M., Kanishcheva M.G., Stepanov S.N. (2022). Observation system resource planning in presence of access control based on volume of resource occupied by traffic flows. T-Comm, vol. 16, no.8, pр. 54-62. (in Russian)
I. Introduction
Our communities have gone through numerous stages of growth and have seen breakthroughs in several socioeconomic areas during the last two decades. Although these advancements have helped to improve our everyday lives, they have also introduced numerous complications into our lives, particularly in terms of safety and security. Security systems are now an integral part of daily life and are used in a variety of public and private settings, including airports, train stations, public parks, offices, and traffic signals. These security measures not only protect people from harm but also detect and document criminal activity. Video surveillance, which successfully enables real-time monitoring and recoding of actions, is thus a crucial part of modern security systems.
With the advent of IoT, smart video surveillance systems (VSS) are gaining popularity over traditional surveillance systems. Human resources are necessary to continually monitor and watch the numerous actions in these traditional surveillance systems, which makes them costly and less trustworthy owing to potential human mistakes. Modern automated and smart surveillance systems, on the other hand, require little human intervention, and their high-quality cameras are integrated with various types of smart sensors to provide better applications such as motion detection, event prediction, real-time tracking, facial recognition, and so on. Millions of cameras are deployed in various places throughout the world to enhance these surveillance systems in order to provide security and aid in society management [1,2].
Cameras in video surveillance systems capture and collect large volumes of data every day, which is then analyzed to extract various sorts of information. The majority of the data gathered is video feeds, which are recorded by several smart cameras to monitor criminal actions in detail and enabling monitoring over a large region [3]. Hence, security and surveillance are crucial in smart environments and account for a significant portion of the Internet of Things, distributed video-based surveillance has a vital role to play in gathering visual data (videos and images) for monitoring and surveillance applications linked to IoT network systems.
Smart or intelligent surveillance combines sensing (smart meters, actuators, and sensors) with video processing to deliver precise and accurate findings based on heterogeneous data streams. Consider the following scenario: smart meters or sensors installed on railway lines indicate that a train is speeding, which may be quickly equated or cross-verified with video information. This will be a key component of the smart IoT ecosystem, and 5G provides frameworks for combining two or more heterogeneous large data streams for accurate decision making [4,5].
However, while collecting such a large amount of heterogeneous big data, resource management and scheduling are the key challenges [6-9]. Supporting video via wireless networks takes a substantial amount of resources. Such services are anticipated to be in high demand and given priority. Research is thus required to determine how to effectively use the existing radio resources for the transmission of video content. Therefore, a framework for taking wireless video transmission into account holistically is needed, notably in surveillance and monitoring systems.
The problem can be resolved using teletraffic theory models and methodologies as well as the built-in capabilities of the mechanisms for controlling the process of servicing communication sessions in contemporary wireless multiservice access nodes. Further, the network needs to evolve from a physical connecting medium into an intelligent technological object capable of offering
an infinite number of services. There are many research works dedicated to address this issue but these works are either partial or do not consider various essential characteristics required for servicing of heterogenous traffic. For example, research conducted by [10], the process ofjoint servicing in the wireless access node of the application flow of LTE subscribers and IoT devices was considered, but the possibilities ofjoint servicing of several multimedia traffic streams and elastic properties of data traffic were not considered. In the works of [11] and [12], the process of sharing an access node resource when servicing multiple elastic traffic flows was investigated, but the possibility of participating in the servicing of real-time traffic flows was not considered. In other works, the task of creating conditions for differentiated servicing of multimedia traffic in a wireless access node was only partially solved. Thus, it seems relevant to study the possibilities of creating conditions for differentiated service in a wireless node of heterogeneous traffic, which is a mixture of several streams of real-time priority traffic and elastic data flow, using access restrictions for all types of communication sessions.
In this context, our work in this paper is dedicated to solve this issue by establishing an analytical framework for modeling the resource sharing process for an operator who intends to design and implement an observation system. This observation system is a WMSN (wireless multimedia sensor networks) implementation that comprises of two types of video cameras. One area is devoted to the recording and production of high-quality HD videos, which are utilized for real-time monitoring. While another type of camera produces low-quality (low-rated) movies that are delivered as files in the system, these videos can be utilized for Videos on Demand (VoD) or video analysis.
II. Heterogenous Networks
The paradigm is changing as 5G wireless technology and the Internet of Things (IoT) are flourishing, and more smarter networks are being implemented. These intelligent networks are decision-making systems that emerged from human intervention systems (for example, communication systems between individuals) and now dominate human activity in a variety of industries, including healthcare, finance, security, and surveillance [13].
However, it is a huge paradigm shift for the operators, who are exploring new ways to provide differentiated services and handle massive volumes of data originating from these networks. Heterogeneous networks are networks that deal with diverse and complex data Hows "Big data" and operate in a way that provides improved capacity and coverage over traditional homogeneous networks. Because of the existence of varied heterogenous data (e.g. Elastic and Real-time) scheduling in these networks is considered complicated yet critical as compared to homogenous networks [14-20]. Thus, development of an ideal scheduler to supervise complex diverse traffic and efficient resource management is very crucial.
Scheduling scenarios for heterogeneous networks are different from those for homogeneous networks because they must handle heterogeneous traffic. To ensure optimal throughput and QoS provision for heterogeneous networks, this is a crucial and difficult issue.
This challenge is brought on by the various demands for various heterogeneous traffic flows. Although the high-band cellular systems in use today (3G/4G) can manage heterogeneous traffic, the exponential rise of data over time has made this challenging. Furthermore, it is proving challenging for these technologies (LTE, LTE-A) to satisfy both individual expectations and the QoS requirements of new emerging applications.
Heterogeneous networks favour the SC-FDMA approach for uplink transmission since it has a larger PRAP. However, due to the heterogeneity in traffic created by the latest applications, particularly with the rise of IoT and smart environments, uplink scheduling in heterogeneous networks remains a difficult problem. The two traffic challenges in homogeneous networks (Elastic traffic and traffic with QoS requirements) have been widely explored, while heterogeneous traffic (e.g. a combination of Elastic and Real-time) within heterogeneous networks is currently being evaluated and researched. Scheduling is challenging in such networks for three reasons: fluctuating channel quality over time owing to stochastic fading effects, "starvation" of poor users (if only users with excellent channel quality are picked), and dealing with applications withvarying QoS needs [21].
Network slicing (NS) procedure is considered one of the key solutions to address the issue of RRM in heterogenous networks, where an underlying physical network is divided in virtual segments known as slices. Although the NS idea was originally presented and used in5G cellular networks, it should be emphasized that this method is equally applicable in LTE and LTE-A networks. In our analysis, which is intended for an observation surveillance system, NS is used to differentiate multiple heterogeneous traffic flows and handle these flows separately through segmentation of the underlying physical network. The technique is performed in such a way that a separate LTE cell with a base station in the center handles the operation of resource sharing.
III. Video Survillance Systems
With the emergence of IoT, smart video surveillance systems (VSS) are overtaking traditional surveillance systems in demand. To improve these VSS and provide security and aid in society management, millions upon millions of cameras are deployed throughout the world's cities [23,23]. Typically, a video surveillance system consists of three primary parts. These include network (analog, digital, or wireless), analytics, and monitoring (cameras, sensors, etc), (behavior analysis, tracking, motion detection, facial recognition etc.). As an illustration, the acquisition component (which aids in obtaining pictures from high- or low-quality cameras), the processing section (or switching part), and the Pan-Tilt-Zoom or PTZ control (its function is to control the positions of the deployed cameras via PTZ functionality).
Analog VSS, digital VSS, and network VSS are the three types of video surveillance systems that may link to analytics (mainframe) in different ways. In analog VSS, coaxial cables are used for shorter-distance transmission of captured and recorded data (images and videos), and optical fibers are utilized for longer-distance transmission. The use of existing LAN and internet infrastructures for the transmission and interchange of digital information (digital image and video files) between devices and analytics is made possible by the IP network used to connect video cameras to a video server in digital VSS [24]. But nowadays wireless sensor networks (WSN) are more recognized for their capabilities.
WSNs are consists of numerous nodes, these nodes are outfitted with sensors (for temperature, pressure, noises, motion, etc.) to monitor their internal operations and collect a variety of information. The information is subsequently transmitted to analytics through BSs. These WANs have been developed further to become WMSNs (wireless multimedia sensor networks), which incorporate a system made up of video cameras and smart sensors
to work together to collect more precise data and also offer realtime visual monitoring. A large video sensor network called WMSN is made up of many video nodes that are outfitted with cameras, sensors, and transceivers to gather both scalar and audiovisual data and transfer it across BSs to analytical centers. As a result, these systems have applicability in a variety of settings, including airports, hospitals, businesses, and public parks etc. to heighten security and safety against potential threats.
But as discussed earlier, the main challenge is RRM and to address this we create an analytical framework for modeling the resource sharing process for an operator aiming to design and use an Observation surveillance system (OSS). Our OSS, which comprises of two different sorts of video cameras, is an implementation of WMSN. High definition videos are produced and captured in one category specifically for real-time monitoring.
While some cameras create low-quality (low-rated) movies that may be utilized for Videos on Demand (VoD) or video analysis, others produce that are communicated as files in the OSS. The video footage gathered by the cameras can be either low definition (poor quality) videos that can be used for Video on Demand (VoD) and are sent as files over the network, or 720p HD videos with a resolution of 1280 x 720 pixels (px) that can be used for real-time surveillance. The HD videos are encoded using a Main profile encoder at 3.1 Level and run at 30 frames per second while using the H.264/AVC codec. These videos' bitrates, which vary from 500 kbps to 1 MB/s, provide the optimal bandwidth and quality trade-offs. The overall amount of bandwidth needed is equal to the total amount of data that each equipped camera transmits.
We assume a single LTE cell with a base station (BS) in the cell's center point that is in charge of managing the resource sharing procedure in order to implement OSS. For serving traffic streams produced by HD security cameras and low-quality cameras computed in units of the lowest granularity, network slicing (NS) provides the number of radio resources of an LTE cell that are accessible in the uplink (UL) direction [25].
A single common underlying infrastructure may be physically divided using NS methods to create independent, specialized network components for a variety of services and needs. The traffic generated by the two types of UEs (Video cameras) is categorized based on quality, bitrate, QoS requirements, and intended use. The model also uses an access control mechanism to provide circumstances for differential servicing of incoming data. There are two classes of traffic flows; one is referred as "Light" traffic (video data from low-quality video camera equipment), while other class of traffic flows is referred to as "Heavy" traffic (real-time high-quality video traffic). Figure 1 depicts the basic representation of OSS.
IV. Description Of Mathematical Model
The total available resources for the given cell in uplink (UP) are given by network slicing (NS) for servicing the real-time traffic arriving from surveillance video cameras and elastic traffic of telemetric devices. For the convenience of modeling, we introduce the concept of a virtual resource unit "r.u.". The information transfer rate provided by one resource unit is assumed to be equal to the minimum information transfer rate requirement from coming requests. Then parameter v is used to denote total number of resources expressed in resource units, while transmission speed offered by one unit is denoted by r. Basic structure of the model and the incoming real-time and elastic traffic flows which are to serviced is shown in Figure 1.
Figure 1. The functional model ofresource sharing between "Heavy" and "Light" video traffic streams
If we consider real-time traffic originating from the surveillance cameras is varying in quality, then the corresponding realtime sessions created are varying by volume.
Based on this we consider having n types of sessions, where real-time sessions of type k arrive according to poissonian flow of intensity, denoted by Ak. In order to get service, each session needs bk resource units and each session occupies assigned resource for random amount of time having exponential distribution with parameter nk while k = \,...,n.
The requests for elastic data (e.g. files) transmission arrive according to poissonian flow with an intensity of Ad. Let's assume i denotes the number resource units engaged by real-time traffic and the number of files that are being serviced in the system is denoted by d. If a request for file transmission follows i + d <v, then it's accepted for servicing. Let's suppose that volume of file to be transmitted is exponentially distributed with mean value of Fbits. Then it can be assumed, that servicing time of one request for data transmission with at least one resource unit has an exponential distribution with parameter = r /F. Real-time traffic requests have priority over elastic traffic requests. On the arrival of this type requests, if there are no available resource units, data transmission rate is reduced (if possible) to release required number resource units. The number of resource units required for servicing a request for data transmission can be reduced only by one unit.
To create conditions for differentiated servicing access control for resource sharing is used. For kth flow of real-time traffic sessions, parameter vk denotes the maximum number of allowed resource units for its simultaneous usage. If vk = v, then incoming sessions of kth flow can use any free available resource units for servicing. Similarly, for elastic traffic sessions vd denotes the maximum number of allowed resource units that can be used by it at the same time. lfvd = v, then incoming sessions of kth flow of elastic traffic can use any free available resource units for servicing. Further, ck represents maximum number of allowed real-time sessions of kth flow that can be served simultaneously. The mathematical model for resource sharing between real-time and elastic data flows is shown in Figure 2.
Figure 2. The mathematical model for resource sharing between real-time and elastic traffic flows
Let's denote by ik(t) the number of real-time sessions for kth flow being on service at time t and by d(t) we denote number of sessions of elastic traffic being on service at time t. The dynamic changes in the states of the model are described by Markov process r(t) = (ij(t),...,in(t),d(t)), defined on the finite set of
model's states S. Let's denote by (ij,...,in,d) the state of r(t).
The vector (i1,...,in,d) belongs to S when ik,k = \,...,n,dvar-
iesasfollows: 0<ik <ck, k = ],...,n; d<vd; ilbl+... +inbn +d<v.
Also, let's denote by p(iin,d) the value of stationary probability of state (i,..., in, d) gS.
Let's suppose that the values of p(iin,d) are known
and define the main performance measures of the model. For kth flow of real time sessions they are: nk - the ratio of lost sessions, mk - the mean number of occupied r.u. For elastic sessions they are: nd - the ratio of lost sessions, yd - the mean number of elastic sessions that are in service, Id - intensity of the termination of the elastic data transfer sessions, kd - average number of resource units used when transmitting elastic data, Td - average service time of elastic session. Formal definitions of introduced performance measures are as follows:
nk =
m, = k
nd =
h =
Z p (h'-ind )>
{(V-n >d ) eS\'kbk +bk >vkor 1+d+bk >v}
S p (h>-j„d) 1kbk '
(11,...,1n,d) eS
Z p (j\,-,in d )'
|,d) eS|d+1>Vd or i+d+1>vj
Z p(h>--jn>d)(v-i)ud ,
{(i1,...,in, d) e S| d > 0}
yd = Z p Qu-Jn d d >
(i1,...,in,d) eS
K =
Id
yd Vd
Td =
1A.
Àd )
To find the introduced performance measures it is necessary to compose and solve the system of state equations. The analysis of numerical methods for solving, showed that the most effective way to solve it is the Gauss-Seidel algorithm.
The numerical data produced by the model shows that when heterogeneous traffic is served by a shared resource, it is uncontrollably redistributed in favor of "Light" traffic (in our model it has been considered as elastic traffic). If conditions are made for differentiated servicing of incoming information flows, the negative effects of this phenomena may be eliminated. Utilizing either static or dynamic slicing will accomplish this. In a static scenario, the available resource is partitioned into slices by incoming streams. The size of the slices is determined by the needs of the communication sessions in relation to the quality of service parameters. For a dynamic scenario, the allocation of a resource in common pool to an incoming request depends on the load of the resource by the communication sessions of the flow under consideration. If the set limit is exceeded, the incoming application is refused, even if a free resource is available.
To evaluate the characteristics of real-time session servicing in a separate slice, it is enough to put Ad = 0. The values of characteristics can be calculated by an effective recursive algorithm based on convolution of vectors of individual distributions of the probability of resource occupancy [26].
Let us introduce vectors Pk with components Pk = (Pk (0), Pk( 1),..Pk(ckbk )), where Pk(/'), is the unnormalized probability of i resource units being occupied when the resource is individually used by requests of the k4 flow. Let's put Pk(0) = 1.
Then
P (0 =
ik!
i=\bk' i = ck'
otherwise
Let us denote by p(i) the stationary probability of occupancy of i resource units of the node to service incoming requests. From the multiplicativity property and the definition of p(i) it follows
pi>=N P
]<o-
Where N is the normalization constant.
Denoting by P(I) = P1 ® P1 ® ,..., ® Pi is the vector resulting from the convolution of the first l vectors Pk= \,...,l. When performing the last convolution, the unnormalized values of the characteristics of the nth stream are found as nn = nn r + nn a, and mn = anbn(\- nn), where
n = V
i=v-b +1
pay,
n = P(cb) X Pn~l\i-cnbn).
na nv n n v n n
The true values of nn, mn andp(i) are found after calculating the normalization constant N = £V=o P© and subsequent normalization. To estimate the characteristics of all n flows, it is necessary to repeat the convolution of the vectors of individual distributions n times. The total number of convolution are n(n-l).
This value can be reduced to 4n - 6 if convolutions are carried out in a certain sequence and the results of intermediate calculations are memorized.
To evaluate the service characteristics of elastic traffic transmission sessions in a separate slice, we can use the generalized access node model, by putting A± = ... An = 0. It shows that file servicing indicators are estimated using the characteristics of the M/ M /1 + w model, where 1+ w isthe maximum possible number of files that can be transferred simultaneously in a slice.
V. Access Node Information Transmission Resource Planning
The model of a node with limited access and algorithms for evaluating the service indicators of incoming applications obtained in the earlier section can be used to analyze the dependence of the characteristics of the quality of service of incoming applications on the parameters of emerging information flows and conditions of access of applications to the occupation of the resource and to select the access level in order to create an advantage in the quality of service for individual information flows. Here is an example of the dependence of the service characteristics of incoming requests for establishing communication sessions on access conditions.
Figure 3, shows the values of the percentage of lost requests depending on the access value found for the node model with the following input parameter values: v = 200 r.u.; n = 4;bk = k r.u.;
ak =
v p nbt
Erl.; P = 1, vk = d, k= 1,2, 3,4; d =50 .. . 100 r.u.
The value of P - loading a resource unit with potential traffic ro = (a1 hi + a2 b2 +a3 b3 +a4 64 )/v. The values of the characteristics are calculated using the convolution algorithm.
65 7(1 75 80 Access restriction, d
Figure 3. Dependence ofreal-time traffic session losses on access conditions
The findings of the computations shown in Figure 4 reveal that for small d, the values of Kk depend non-linearly on the change in d (because of the integer character bk). It can be seen, as value of d increases, there is decrease in the request losses and quickly converge to the limit values obtained at d = v. Simultaneously, the difference in the values of losses in the presence of an access restriction and without the use of an access restriction can reach several times. Thus, by reducing or increasing the access of applications to the resource, it is possible to reduce or increase their service advantage. The marked property will be further used to create a service advantage for real-time traffic.
Here is another example of using access restrictions to create conditions for differentiated servicing of incoming flows of heterogeneous traffic. Let's consider the access node model with following parameters:
v = 200 r.u.; n =3; b = 1 r.u.; b2 = 10 r.u.; b3 = 20 r.u.; Erl; p=l; vk = v r.u.; k= 1, 2, 3.
vp
ak= -
nb.
The value of p - is loading a unit of resource.
Depending on the requirements for individual flow transfer rates and the amount of resource utilized to serve incoming requests, Figure 4 depicts the percentage of lost requests. The given data shows that in the absence of access restriction, the loss values are strongly influenced by bk. It shows higher the bk value, the higher the Kk losses. It is concluded that 282 r.u. is required to provide QoS with loss not exceeding 0.05. By varying the access value, it is possible to equalize the losses Kk We choose access restrictions based on the following expressions: vj =|^v x 0.27J , v2 x0.47j , v3 = v .
The rest of the model parameters remained unchanged. Figure 5, depicts the percentage of requests that are lost based on the requirements for particular flow transfer rates and the amount of resource utilized to handle incoming requests.
From the presented data, we can witness that the presence of access restrictions equalizes the values of Kk losses, thereby creating the same conditions for servicing incoming orders. We also note the non-linear character of the dependence of Kk on v. This attribute is associated with the usage of the access restriction and the value bk. To ensure QoS with losses not exceeding 0.05, smaller number of channels i.e. 278 r.u. is required.
The numerical study revealed that when servicing data in realtime sessions, the use of dynamic slicing reduces losses by 1-2% in differentiated services aimed at equalizing session losses on a fixed amount of resource, and by 2-5% in the amount of resource required to provide the required level of session losses, when compared to static slicing for the same purposes. The efficiency of dynamic slicing is substantially higher when providing data on the basis of elastic traffic sessions, i.e. when employing the Processor Sharing discipline. Its use can reduce losses by up to two times in the case of differentiated services aimed at leveling session losses on a fixed amount of resource, and it can reduce the requirement for the volume of a resource that provides the required level of session losses by 10-15% when compared to static slicing.
Figure 5. Equalization of real-time traffic session losses based on access restriction
Comparison of static and dynamic scenarios is considered for the model of serving two streams of "heavy" and "light" traffic sessions with the following parameters: v = 200 r.u.; transmission rate that is provided by one resource unit is 1 r.u. = r= 100 kbit/s; b1 = 10 r.u.; F = 100 kbit; 1/^ = 10 s; M[id = 1 s; v1 = v; h = 200^x / 2b1 requests per unit of time; Ad = 200/ud/ 2 requests per unit of time. Thus, one session of "heavy" traffic requires a transmission rate of 1 Mbit/s, and one session of "heavy" traffic is served at a rate of 1 r.u = r = 100 kbit/s. The unit of time is the average time of file transfer using the capabilities of one unit of the resource.
So, if we even out the losses at the 0.03 level using dynamic slicing, we may determine how much resource is required by utilizing elaborated in model and derived methods of its performance measures estimate. Figure 6, shows the results of the loss calculation. When access is limited to sessions of "light" traffic at the level of v d = 61 r.u. on the shared resource of size v = 218 r.u., the losses become equivalent to around 0.03. Using the static situation, serving real-time traffic costs 160 r.u. while serving elastic traffic requires 97 r.u. Thus, using access restriction leads in a savings of 39 r.u., or 15%.
Figure 6. Using dynamic slicing to equalize the losses of sessions of"heavy" and "light" traffic at the level of0.03
Figure 4. Dependence of the loss ofreal-time traffic transmission sessions on the transmission speed requirements
VI. Conclusion
In contrast to previously developed models, a generalized model of the establishment and maintenance of communication session flows of a wireless access node has been created with respect to an observation surveillance system. This model allows for the joint influence of the primary significant factors determining the joint transmission of traffic of modern communication applications. There are several of them, including: the presence of priority for real-time traffic; using the Processor Sharing discipline when transmitting elastic traffic; access restrictions for all types of communication sessions depending on the resource loading by the communication sessions of the stream in question.
A scenario for the dynamic distribution of the access node resource between incoming sessions has been developed using the designed model, allowing for the creation of conditions for differentiated maintenance of heterogeneous traffic flows based on access restrictions based on the resource loading by the communication sessions of the concerned stream.
The conducted numerical study shows that the service of heterogeneous traffic by a common resource leads to its uncontrolled redistribution in favor of "light" traffic. It is possible to eliminate the negative consequences of this phenomenon if conditions are created for differentiated servicing of incoming information flows. This may be accomplished by either the static or dynamic slicing options. In the first case, the available resource is divided between the incoming flows in a certain proportion, depending on the requirements of the communication sessions to the QoS indicators. While in the case second, the allocation of the resource is carried out on a dynamic basis and depends on its load by each of the served threads.
References
1. Rost P. Mannweiler C, Michalopoulos DS, et al. (2017). Network slicing to enable scalability and flexibility in 5G mobile networks. IEEE Communications Magazine. Vol. 55. No.5. P. 72-79.
2. D. Cumming, and S. Johan (2015). Cameras Tracking Shoppers: The Economics of Retail Video Surveillance. Eurasian Business Review. Vol. 5. No.2. P. 235-257.
3. Y. Tang, B. Ma, and H. Yan (2013). Intelligent Video Surveillance System for Elderly People Living Alone Based on ODVS. Advances in Internet of Things. Vol.3. No. 2. P. 44-52.
4. W. Zhu, P. Cui, Z. Wang, and G. Hua (2015). Multimedia Big Data Computing. IEEE Multimedia. Vol. 22. No.2. P. 95-105.
5. D. Che, M. Szafran, and Z. Peng (2013). From Big Data to Big Data Mining: Challenges, Issues, and Opportunities. Proc. of the 18th International Conference on Database Systems for Advanced Applications. P. 1-15.
6. S. N. Stepanov, U. M. Andrabi, M. S. Stepanov and J. Ndayikunda (2020). Reservation Based Joint Servicing of Real Time and Batched Traffic in Inter Satellite Link. Systems of Signals Generating and Processing in the Field of on Board Communications, Moscow, Russia. P. 1-5.
7. S. Stepanov, M. Stepanov , A. Tsogbadrakh , J. Ndayikunda, and U. Andrabi (2019). Resource Allocation and Sharing for Transmission ofBatchedNB IoT Traffic over 3GPP LTE. Conference of Open Innovations Association, FRUCT. P. 422-429.
8. M. A. Umer, and S. N. Stepanov (2020). Investigation of Cellular Network Resource Division Procedures for the Joint Servicing of Real-Time Multiservice Traffic and Elastic IoT Traffic. Synchroinfo Journal. Vol. 6. No.l. P. 7-10.
9. S.N. Stepanov, M.S. Stepanov, U.M. Andrabi, S.N., and J. Ndayikunda (2020). The Analysis of Resource Sharing for Heterogenous
Traffic Streams over 3GPP LTE with NB-IoT Functionality. Distributed Computing and Computer Networks. DCCN 2020. Lecture Notes Computer Science, Springer, Cham. Vol.12563. P. 422-435.
10. V. Begishev, V. Petrov, A. Samoylov, D. Moltchanov, S. Andreev, Y. Koucheryavy, and K. Samouylov (2018). Resource Allocation and Sharing for Heterogeneous Data Collection over Conventional 3GPP LTE and Emerging NB-IoT Technologies. Computer Communications. Vol. 120. No.2. P. 93-101.
11. J. Roberts, U, Mocci and J. Virtamo (1996). Broadband Network Traffic: Performance Evaluation and Design Of Broadband MultiserviceNetworks. Springer,1996. 584 p.
12. T. Bonald (2007). Insensitive traffic models for communication networks. Discrete Event Dynamic Systems. Vol.17. No.3. P. 405-421.
13. O. Elharrouss, N. Almaadeed, S. Al-Maadeed (2021). Review of Video Surveillance Systems. J Vis Commun. Image Represent. Vol.77. 103116p.
14. M. S. Stepanov, S. N. Stepanov, U. Andrabi, D. Petrov, and J. Ndayikunda (2022). The Increasing of Resource Sharing Efficiency in Network Slicing Implementation. Distributed Computing and Computer Networks. DCCN 2021. Lecture Notes Computer Science, Springer, Cham. Vol. 1552.P. 18-35.
15. U. M. Andrabi, and S. N. Stepanov (2021). The Model of Conjoint Servicing of Real-Time Traffic of Surveillance Cameras and Elastic Traffic Devices with Access Control. 2nd International Informatics and Software Engineering Conference (IISEC). P. 1-6.
16. U. M. Andrabi, S. N. Stepanov, M. S. Stepanov, M. G. Kanishcheva and F. X. Habinshuti (2021). The Model of Conjoint Servicing of Real Time and Elastic Traffic Streams Through Processor Sharing (PS) Discipline with Access Control. International Conference Engineering and Telecommunication (En&T). P. 1-5.
17. U. M. Andrabi, S.N. Stepanov., J. Ndayikunda , and M.G. Kanishcheva (2020). Cellular Network Resource Distribution Methods for The Joint Servicing of Real-Time Multiservice Traffic And Grouped IoT. T-Comm. Vol.14. No.10. P. 61-69.
18. U.M. Andrabi and F.X. Habinshuti (2021). Radio Resource Management For Shared Servicing of Real-Time Multiservice and Elastic IOT Traffic Streams. International Journal of Electrical, Electronics and Data Communication 167 (IJEEDC). Vol.9.1.12. P. 1-6.
19. S. N. Stepanov, A. V. Korobkina, A. O. Volkov, E. E. Malikova, A. E. Panov (2021). The Analysis of Traffic Balancing for Data Centers Serving Requests of LEO Mobile Satellite Systems. 2021 Systems of Signals Generating and Processing in the Field of on Board Communications,pp. 1-6.
20. U. M. Andrabi and M.S. Stepanov (2022). Collective Servicing of Heterogenous Traffic Streams over 3GPP LTE Network and Application of Access Control. T-Comm. Vol.16. No.3. P. 43-49.
21. N. Abu-Ali, A. M. Taha, M. Salah and H. Hassanein (2014). Uplink Scheduling in LTE and LTE-Advanced: Tutorial, Survey and Evaluation Framework. IEEE Communications Surveys & Tutorials. Vol.16. No.3. P.1239-1265.
22. Rost P. Mannweiler C, Michalopoulos DS, et al. (2017). Network slicing to enable scalability and flexibility in 5G mobile networks. IEEE Communications Magazine. Vol.55. No.5. P. 72-79.
23. D. Cumming, and S. Johan (2015). Cameras Tracking Shoppers: The Economics of Retail Video Surveillance. Eurasian Business Review. Vol.5. No.2. P. 235-257.
24. J. C. SanMiguel, C. Micheloni, K. Shoop, G. L. Foresti and A. Cavallaro (2014). Self-Reconfigurable Smart Camera Networks. Computer. Vol.47. No.5. P. 67-73.
25. Y. Yu and J. Wang (2018). Uplink Resource Allocation for Narrowband Internet of Things (NB-IoT) Cellular Networks. Asia-Pacific Signal and Information Processing Association Annual Summit and Conference. P. 446-471.
26. V.B.Iversen (1987). The exact evaluation of multi-service loss systemwith access control. Teleteknik. Vol.31. No.2. P. 56-61.
ПЛАНИРОВАНИЕ РЕСУРСА СИСТЕМ ТЕЛЕНАБЛЮДЕНИЯ ПРИ НАЛИЧИИ КОНТРОЛЯ ДОСТУПА, ОСНОВАННОГО НА ОБЪЕМЕ РЕСУРСА ЗАНЯТОГО ПОТОКАМИ ТРАФИКА
Андраби Умэр Мукхтар, Московский физико-технический институт (государственный университет), Москва, Россия.
Канищева Маргарита, Московский Технический Университет Связи и Информатики (МТУСИ), Москва, Россия. Степанов Сергей Николаевич, Московский Технический Университет Связи и Информатики (МТУСИ), Москва, Россия
Аннотация
Благодаря недавним достижениям в области сотовых технологий в сетях связи обслуживаются информационные потоки с самыми разнообразными требованиями к скорости передачи и показателям качества обслуживания. Особенно ярко перечисленные свойства проявляют себя в интеллектуальных сетях, в частности при использовании технологии IoT. Регулирование и качественная передача таких огромных объемов данных вызывает серьезные опасения у виртуальных операторов, представленных на рынке связи, и использующих инфраструктуру сотовых сетей для организации сервисов видеонаблюдений для своих клиентов. К сожалению, не существует единой методологии, позволяющей реализовать качественное обслуживания поступающих потоков трафика и распределение ресурса передачи информации в гетерогенных сетях. Одним из способов решения перечисленных задач является математическое моделирование, учитывающее особенности транспортных потоков, возникающих и принимаемых для обслуживания. Чтобы решить эту проблему, в работе построена стратегия распределения ресурса при совместном использовании ресурса изолированной соты сети LTE потоками гетерогенного трафика. Источниками трафика являются видеокамеры оператора систем наблюдений, использующего инфраструктуру сети сотовой связи для передачи полученных данных в аналитические центры. Процесс совместного использования ресурса анализируется с помощью математической модели, в которой учитывается наличие приоритета у трафика реального времени, свойство эластичности при передаче файлов и ограничение доступа для всех потоков, зависящее от объема ресурса, занятого каждым потоком. Построены алгоритмы оценки характеристик обслуживания заявок и исследована зависимость характеристик от ограничения доступав. С использованием разработанной модели разработан сценарий динамического распределения ресурса узла доступа между входящими сеансами, позволяющий создавать условия для дифференцированного обслуживания разнородных потоков трафика на основе ограничений доступа, основанных на загрузке ресурса каждым из рассматриваемых потоков.
Ключевые слова: управление радиоресурсами (RRM), потоки видеотрафика, IoT, сегментирование сети (NS), оператор систем наблюдений, контроль доступа.
Литература
1. Rost P. Mannweiler C, Michalopoulos DS, et al. Network slicing to enable scalability and flexibility in 5G mobile networks // IEEE Communications Magazine. 2017. Vol. 55. No.5, pp. 72-79.
2. D. Cumming, and S. Johan. Cameras Tracking Shoppers: The Economics of Retail Video Surveillance // Eurasian Business Review. 2015. Vol. 5. No.2, pp. 235-257.
3. Y. Tang, B. Ma, and H. Yan. Intelligent Video Surveillance System for Elderly People Living Alone Based on ODVS // Advances in Internet of Things. 2013. Vol.3. No.2, pp. 44-52.
4. W. Zhu, P. Cui, Z. Wang, and G. Hua. Multimedia Big Data Computing // IEEE Multimedia. 2015. Vol. 22. No.2, pp. 95-105.
5. D. Che, M. Szafran, and Z. Peng. From Big Data to Big Data Mining: Challenges, Issues, and Opportunities // Proc. of the 18th International Conference on Database Systems for Advanced Applications. 2013, pp. 1-15.
6. S. N. Stepanov, U. M. Andrabi, M. S. Stepanov and J. Ndayikunda. Reservation Based Joint Servicing of Real Time and Batched Traffic in Inter Satellite Link // Systems of Signals Generating and Processing in the Field of on Board Communications, Moscow, Russia. 2020, pp. I-5.
7. S. Stepanov, M. Stepanov, A Tsogbadrakh, J. Ndayikunda and U. Andrabi. Resource Allocation and Sharing for Transmission of Batched NB IoT Traffic over 3GPP LTE // Conference of Open Innovations Association, FRUCT. 2019. P. 422-429.
8. M. A. Umer, and S. N. Stepanov. Investigation of Cellular Network Resource Division Procedures for the Joint Servicing of Real-Time Multiservice Traffic and Elastic IoT Traffic // Synchroinfo Journal. 2020. Vol. 6. No.I, pp. 7-10.
9. S.N. Stepanov, M.S. Stepanov, U.M. Andrabi, S.N., and J. Ndayikunda. The Analysis of Resource Sharing for Heterogenous Traffic Streams over 3GPP LTE with NB-IoT Functionality // Distributed Computing and Computer Networks. DCCN 2020. Lecture Notes Computer Science, Springer, Cham. 2020. Vol. 12563, pp. 422-435.
7TT
10. V. Begishev, V. Petrov, A. Samoylov, D. Moltchanov, S. Andreev, Y. Koucheryavy, and K. Samouylov. Resource Allocation and Sharing for Heterogeneous Data Collection over Conventional 3GPP LTE and Emerging NB-IoT Technologies // Computer Communications. 2018. Vol. 120. No.2, pp. 93-101.
11. J. Roberts, U, Mocci and J. Virtamo. Broadband Network Traffic: Performance Evaluation and Design Of Broadband Multiservice Networks. Springer, 1996. 584 p.
12. T. Bonald. Insensitive traffic models for communication networks // Discrete Event Dynamic Systems. 2007. Vol. 17. No.3, pp. 405421.
13. O. Elharrouss, N. Almaadeed, S. Al-Maadeed. Review of Video Surveillance Systems // J Vis Commun. Image Represent. 2021. Vol. 77. 103116 p.
14. M.S. Stepanov, S.N. Stepanov, U. Andrabi, D. Petrov, and J. Ndayikunda. The Increasing of Resource Sharing Efficiency in Network Slicing Implementation // Distributed Computing and Computer Networks. DCCN 2021. Lecture Notes Computer Science, Springer, Cham. 2022. Vol. 1552, pp. 18-35.
15. U. M. Andrabi, and S. N. Stepanov. The Model of Conjoint Servicing of Real-Time Traffic of Surveillance Cameras and Elastic Traffic Devices with Access Control // 2nd International Informatics and Software Engineering Conference (IISEC). 2021, pp. 1-6.
16. U. M. Andrabi, S. N. Stepanov, M. S. Stepanov, M. G. Kanishcheva and F. X. Habinshuti. The Model of Conjoint Servicing of Real Time and Elastic Traffic Streams Through Processor Sharing (PS) Discipline with Access Control // International Conference Engineering and Telecommunication (En&T). 2021, pp. 1-5.
17. U. M. Andrabi, S.N. Stepanov, J. Ndayikunda and M.G. Kanishcheva. Cellular Network Resource Distribution Methods for The Joint Servicing of Real-Time Multiservice Traffic And Grouped IoT // T-Comm. 2020. Vol.14. No.10, pp. 61-69.
18. U.M. Andrabi and F.X. Habinshuti. Radio Resource Management For Shared Servicing of Real-Time Multiservice and Elastic IOT Traffic Streams // International Journal of Electrical, Electronics and Data Communication. 167 (IJEEDC). 2021. Vol.9. I.I2, pp. 1-6.
19. S.N. Stepanov, A.V. Korobkina, A.O. Volkov, E.E. Malikova, A.E. Panov. The Analysis of Traffic Balancing for Data Centers Serving Requests of LEO Mobile Satellite Systems // 2021 Systems of Signals Generating and Processing in the Field of on Board Communications, 2021, pp. 1-6.
20. U.M. Andrabi and M.S. Stepanov. Collective Servicing of Heterogenous Traffic Streams over 3GPP LTE Network and Application of Access Control/ // T-Comm. 2022. Vol. 16. No.3, pp. 43-49.
21. N. Abu-Ali, A. M. Taha, M. Salah and H. Hassanein. Uplink Scheduling in LTE and LTE-Advanced: Tutorial, Survey and Evaluation Framework // IEEE Communications Surveys & Tutorials. 2014. Vol.16. No.3, pp. 1239-1265.
22. Rost P. Mannweiler C, Michalopoulos DS, et al. Network slicing to enable scalability and flexibility in 5G mobile networks // IEEE Communications Magazine. 2017. Vol.55. No.5, pp. 72-79.
23. D. Cumming, and S. Johan. Cameras Tracking Shoppers: The Economics of Retail Video Surveillance // Eurasian Business Review.-2015. Vol.5. No.2, pp. 235-257.
24. J. C. SanMiguel, C. Micheloni, K. Shoop, G. L. Foresti and A. Cavallaro. Self-Reconfigurable Smart Camera Networks // Computer. 2014. Vol.47. No.5, pp. 67-73.
25. Y. Yu and J. Wang. Uplink Resource Allocation for Narrowband Internet of Things (NB-IoT) Cellular Networks // Asia-Pacific Signal and Information Processing Association Annual Summit and Conference. 2018, pp. 446-471.
26. V.B. Iversen.The exact evaluation of multi-service loss system with access control // Teleteknik. 1987. Vol.31. No.2, pp. 56-61.