KR102209576B1 - The method of link scheduling for data processing and the apparatus thereof - Google Patents

Abstract

The present invention relates to a link scheduling method and apparatus used for 5G mobile communication in a millimeter wave (mmWave) band. The present invention can solve a complex MWIS (Maximum Weight Independent Set) problem in real time and calculate a backhaul link scheduling result in a relatively short time.

Description

TECHNICAL FIELD The method of link scheduling for data processing and the apparatus thereof

The present invention relates to a link scheduling method and apparatus for data processing, and more particularly, to a link scheduling method and apparatus used for 5G mobile communication in a millimeter wave (mmWave) band.

New business requirements and emerging technologies such as the Internet of Things (IoT) are transforming manufacturing companies' environments into smart manufacturing. In a smart manufacturing system, various IoT-based manufacturing equipment generates, processes, and exchanges large-scale sensory information in an industrial IoT network through context recognition.

This accumulated data requires greater bandwidth and efficiently reconfigurable routing protocols to connect network IoT devices for real-time processing.

There are currently several standards for wireless industrial communications such as WirelessHart, ZigBee, ISA 100 and WiFi. Existing industrial wireless systems provide relatively low data rates for sensor information and traffic control. This type of technology can be used for the last mile of wireless sensor networks, especially mobile manufacturing equipment in smart manufacturing systems.

Wired products generally support higher speeds, but wiring is usually expensive in large manufacturing sites. To solve this problem, the enormous amount of bandwidth available in the millimeter wave (hereinafter referred to as mmWave) 60 GHz spectrum enables large data capacity and high-speed data transmission methods such as the wireless backhaul of smart manufacturing.

It is known that gigabit transmission per second can be obtained in the mmWave band of about 60 GHz. Directional antennas (high antenna gain) are mainly used in this band, which allows for better spatial reuse while lowering the probability of interference due to high signal attenuation in the mmWave band.

Referring to Figure 1, Edge Gateways (EG) aggregate local IoT traffic and be routed through a directional link (directional link) between the Edge Gateway (EG) and the EG and Edge Server (ES). I can. As shown in Fig. 1, the Edge X Foundry layer architecture is composed of four service layers, and each service layer is composed of several embedded microservice modules.

In particular, the scheduling microservice of the support service layer performs backhaul routing, while the mmWave microservice of the device service layer can be used to support communication using a directional antenna (directional communication) through an industrial IoT network.

Locally aggregated real-time traffic requires an efficient scheduling algorithm for the backhaul link to reduce the total average end-to-end transmission delay. Additionally, routing using directional antennas improves network throughput and consumes minimal total energy.

There are various routing protocols based on directional antennas (directional routing), which can traditionally be classified as proactive or reactive approaches. The proactive approach sets routing paths in the local routing table to reduce routing delays. Reactive approach finds routing paths when needed.

However, these techniques are all based on traditional mobile ad hoc routing (MANET) and are limited by end-to-end transmission delays because they do not effectively correct deafness. The hearing impairment problem occurs when a node does not respond to a communication request frame for it, and as a result, the sender transmits more request frames, resulting in a longer contention time. Limits can be exacerbated in intelligent manufacturing systems as real-time control information or sensory data is lost or cannot arrive at the destination edge gateway (EG). Therefore, a technology to improve real-time data transmission is required for smart production systems.

The present invention has been devised to solve the above problem, and proposes a new directional routing link scheduling algorithm for a smart mobile edge in a smart manufacturing system based on a maximum weight independent set (hereinafter referred to as MWIS).

This algorithm is designed to explicitly reduce the transmission delay by using a collision/hearing loss graph to reduce the likelihood of a hearing loss problem between edge gates (EG). In a smart manufacturing system, all edge gates operate under the control of an edge server (ES). Therefore, the edge server maintains a collision-hearing loss graph and determines whether the directional links between the edge gates (EG) interfere with each other or cause a hearing loss problem when scheduled at the same time.

The link scheduling result is calculated by the edge server (ES) and delivered to all EGs through the control channel. The algorithm proposed in the present invention solves the complex MWIS problem in real time by considering edge gateways (EG) in the order of the maximum independent number and allocating time slots to the link having the maximum traffic load.

Considering the characteristics of large-scale real-time backhaul traffic and mmWave technology, the use of a directional antenna can reduce radio interference between EGs and significantly improve packet throughput. In the scheduling process, the algorithm according to an embodiment of the present invention discovers and allocates time slots that maximize space reuse between links.

The technical problems of the present invention are not limited to the technical problems mentioned above, and other technical problems that are not mentioned will be clearly understood by those skilled in the art from the following description.

The link scheduling method according to an embodiment of the present invention is a link scheduling method for data processing in a backhaul network-based network system including a plurality of nodes, and selects any one node to perform scheduling among the plurality of nodes. Step (a); (B) sequentially scheduling a plurality of links connected to the selected node in consideration of collision-hearing loss constraints; (C) repeating the scheduling step (b) from a first time interval by selecting any one of the remaining unselected nodes among the plurality of nodes; (D) searching for a link that satisfies the collision-hearing-loss constraint for each time interval when scheduling of all links is completed once at a time; (E) performing channel utilization by scheduling the searched link in a corresponding time interval; Calculating channel utilization for all time intervals (f); calculating fairness constraints based on the channel utilization (g); (H) selecting a link with the lowest channel utilization among the plurality of links when the fairness constraint is less than a predetermined value; (I) scheduling the link in an additional time interval; (J) repeating steps (f) to (h); And when the fairness constraint is greater than or equal to a predetermined value, terminating the scheduling (k).

In the link scheduling method according to an embodiment of the present invention, in the step of selecting any one node, a node selected as a node to perform the scheduling may be a node having the most links.

In the link scheduling method according to an embodiment of the present invention, in the step of selecting any one node, when all the links have the same number of nodes, any one node may be randomly selected.

In the link scheduling method according to an embodiment of the present invention, the collision-hearing loss constraint may satisfy the following equation.

는 링크 i와 링크 j가 동시에 스케쥴링 되었을 때 난청이 발생하면 1이고 난청이 발생하지 않으면 0,

는 링크 i와 링크 j가 동시에 스케쥴링 되었을 때 충돌이 발생하면 1이고 충돌이 발생하지 않으면 0, d_i(t)={0,1}는 t 시간 슬롯에 스케쥴링 되면 1이고 스케쥴링되지 않으면 0,

을 만족한다.Where i, j are links, t are time slots,

Is 1 if hearing loss occurs when link i and link j are scheduled at the same time, 0 if no hearing loss occurs,

Is 1 if a collision occurs when link i and link j are scheduled at the same time, 0 if no collision occurs, d _i (t)={0,1} is 1 if scheduled in time slot t, and 0 if not scheduled,

Is satisfied.

)는 하기 수학식에 의해 산출될 수 있다.The link scheduling method according to an embodiment of the present invention includes the channel utilization (

) Can be calculated by the following equation.

는 링크 i의 가중치,

는 t 시간 슬롯에서 스케쥴링 되면 1이고, 스케쥴링되지 않으면 0이다.here,

Is the weight of link i,

Is 1 if scheduled in time slot t, and 0 if not scheduled.

In the link scheduling method according to an embodiment of the present invention, the fairness constraint may satisfy the following equation.

는 공정성 임계치,

는 채널 활용도,

는 임의의 수이다.here,

Is the fairness threshold,

Is the channel utilization,

Is an arbitrary number.

In the link scheduling method according to an embodiment of the present invention, the step (i) of scheduling the link in an additional time interval is the step of searching and scheduling a link that satisfies the collision-hearing loss constraint for the additional time interval (l) It may further include.

In the link scheduling method according to an embodiment of the present invention, the network system may be a smart factory.

In the link scheduling method according to an embodiment of the present invention, the backhaul network may use a millimeter wave (mmWave) band.

The link scheduling apparatus according to an embodiment of the present invention is for processing data in a network system based on a backhaul network including a plurality of nodes, and may include a first scheduling unit, a second scheduling unit, and a third scheduling unit.

The first scheduling unit according to an embodiment of the present invention selects any one node to perform scheduling among the plurality of nodes, and considers collision-hearing loss constraints for a plurality of links connected to the selected node. The scheduling may be sequentially performed, and the scheduling may be repeated from the first time interval by selecting any one of the remaining nodes that are not selected from among the plurality of nodes.

When the scheduling of all links is completed once, the second scheduling unit according to an embodiment of the present invention searches for a link that satisfies the collision-hearing-loss constraint for each time interval, and uses the searched link for a corresponding time interval. Channel utilization can be performed by scheduling in.

A third scheduling unit according to an embodiment of the present invention calculates channel utilization for all time intervals, calculates fairness constraints based on the channel utilization,

When the fairness constraint is less than a predetermined value, a link with the lowest channel utilization is selected among the plurality of links, the link is scheduled for an additional time period, and the channel utilization and the fairness constraint calculation and the channel utilization are the lowest. Link selection is repeated, and when the fairness constraint is greater than or equal to a predetermined value, scheduling may be terminated.

Channel utilization according to an embodiment of the present invention may refer to an operation that reduces the free time period as much as possible by further scheduling a link that satisfies the collision-hearing loss constraint for an unscheduled free time period in each time period.

The algorithm proposed in the present invention solves the complex MWIS problem in real time by considering the order of the maximum number of independent edge gateways (EG) and allocating time slots to the link with the maximum traffic load, resulting in the scheduling of the backhaul link in a relatively short time. Calculate

For the transmission fairness of the proposed algorithm according to an embodiment of the present invention, the transmission fairness constraint is numerically analyzed using Jain's fairness index method, and the efficiency of the algorithm includes throughput, delay, packet loss rate and transmission fairness. Was measured by evaluating the performance metric.

Simulation results show that the proposed scheme is superior to the existing mmWave routing scheme.

The effects of the present invention are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a conceptual diagram of a network system according to an embodiment of the present invention.
2 shows an example of a smart manufacturing system backhaul network according to an embodiment of the present invention.
3 is a code of a link scheduling algorithm according to an embodiment of the present invention.
4 shows a plurality of nodes and links that perform link scheduling according to an embodiment of the present invention.
5 and 6 show time slots representing link scheduling according to an embodiment of the present invention.
7 shows a graph of backhaul network throughput versus normalized traffic load.
8 shows the average delay versus normalized traffic load.
9 is a graph showing packet loss rate versus normalized traffic load.
Figure 10 shows the backhaul network throughput versus the normalized traffic load of a large network.
Figure 11 shows the average delay versus normalized traffic load in a large network.

In the present invention, various modifications may be made and various embodiments may be provided, and specific embodiments will be described in detail with reference to the drawings. However, this is not intended to limit the present invention to a specific embodiment, it is to be understood to include all changes, equivalents, and substitutes included in the spirit and scope of the present invention. In describing each drawing, similar reference numerals have been used for similar elements.

Terms such as first, second, A, and B may be used to describe various elements, but the elements should not be limited by the terms. These terms are used only for the purpose of distinguishing one component from another component. For example, without departing from the scope of the present invention, a first component may be referred to as a second component, and similarly, a second component may be referred to as a first component. The term and/or includes a combination of a plurality of related items or any of a plurality of related items.

When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected to or connected to the other component, but other components may exist in the middle. something to do. On the other hand, when a component is referred to as being "directly connected" or "directly connected" to another component, it should be understood that there is no other component in the middle.

The terms used in the present application are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In the present application, terms such as "comprise" or "have" are intended to designate the presence of features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, but one or more other features. It is to be understood that the presence or addition of elements or numbers, steps, actions, components, parts, or combinations thereof, does not preclude in advance.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs. Terms as defined in a commonly used dictionary should be interpreted as having a meaning consistent with the meaning in the context of the related technology, and should not be interpreted as an ideal or excessively formal meaning unless explicitly defined in this application. Does not.

Throughout the specification and claims, when a certain part includes a certain component, it means that other components may be further included rather than excluding other components unless otherwise stated.

Throughout the specification and claims, an edge may refer to a link, and the two terms may be used interchangeably.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Wireless backhaul network system model

First, it is assumed that the wireless backhaul system model supports multi-hop communication between EG, ES and backhaul network in a smart manufacturing environment, and all EG and ES are equipped with switch beam directional antennas. The manufacturing equipment (ME) communicates with the EG by transmitting a communication request frame to the EG located near the ME.

Each EG collects wireless backhaul traffic from the ME, and passes this traffic to the ES using a directional antenna. When backhaul traffic reaches the ES, it passes this traffic to the core network. In addition, it is assumed that the wireless backhaul network supports multi-hop communication between EGs. Therefore, each EG can build a mesh topology and perform multi-hop communication with other EGs except ES. As shown in Figure 1, each ME is connected to an EG, and all EGs are connected to an adjacent EG. The ES connects to the cloud network via a wired backhaul and uses a directional antenna to pass backhaul traffic to the EG.

In general, directional antennas are classified into two types using a switched beam antenna and a phase array antenna. In the present invention, a switched beam antenna widely adopted in various studies such as a wireless backhaul network is used, but is not limited thereto. It is assumed that all EGs are composed of the same number of antenna beams indicated by M. All beams do not overlap to cover all directions and have the same transmission coverage area and transmission power.

However, when the EG transmits radio traffic, only one beam operates at that time to send the transmitted traffic to the desired recipient, and the other beam is blocked due to the use of a single channel. Receiving the wireless backhaul traffic, the recipient EG adjusts the directional beam toward the EG that requested the communication.

In order to transmit wireless backhaul traffic, the EG needs information on adjacent EGs, including beam ID, beam direction, and channel weight. Information on neighboring EGs is periodically collected through a cooperative channel (eg, ISM bands) through a hello message, and the obtained information is stored in the beam table of the EG.

)로 모델링했다. 여기서 N은 모든 EG의 집합을 나타내며

는 모든 방향성 에지의 집합이다. 무선 백홀 네트워크는 한 세트의 데이터 흐름과 한 세트의 직교 채널로 구성된다. EG (n₁)가 무선 백홀 트래픽을 다른 EG (n₂)로 보내는 경우 G의 방향성 에지 e = (n₁, n₂)로 표시된다. EG 세트는 N = {n₁, n₂, .. 여기서, α는 무선 백홀에서 EG의 수를 나타내고, 각각의 n_i는 e로 표시되는 에지를 갖는다. 에지 집합은

= {e₁, e₂, ..., e_β}로 표현되며, 여기서 β는 무선 백홀상의 에지 수를 나타낸다.In one embodiment of the present invention, a wireless backhaul network is directed graph G = (N,

). Where N represents the set of all EGs

Is the set of all directional edges. A wireless backhaul network consists of a set of data flows and a set of orthogonal channels. When EG (n ₁ ) sends wireless backhaul traffic to another EG (n ₂ ), it is denoted by G's directional edge e = (n ₁ , n ₂ ). The EG set is N = {n ₁ , n ₂ , .. where α represents the number of EGs in the wireless backhaul, and each n _i has an edge denoted by e. Edge set

= {e ₁ , e ₂ , ..., e _β }, where β represents the number of edges on the wireless backhaul.

We also defined N(n) as the set of neighboring nodes of node n that are located within the transmission range of node n.

In addition, the edge e r _e of the channel capacity could be derived as follows.

[Equation 1]

W is the channel bandwidth of the wireless backhaul, N ₀ is the noise power spectral density, I is the wireless backhaul interference, P _T is the transmit power of the EG, G _T and G _R are the directional beam gain of the transmitter EG and receiver EG, and λ is the directivity. Is the wavelength of the beam, 4π is the directional antenna reflects 4 sectors, d is the distance between the transmitter EG and the receiver EG.

Interference model

2 shows an example of a smart manufacturing system backhaul network according to an embodiment of the present invention.

= {e₁, e₂, ..., e₁₀}은 EGs의 수 (α)가 5이고 에지의 수 (β)가 10임을 나타낸다. EG에는 4 개의 색인으로 구성된 지향성 안테나가 장착되어 있다. n₁이 n₅로 트래픽을 보내려고 하면 n₂와 n₄는 대상 빔이 차단되어 통신 할 수 없다. 그러나 n₁이 n₅로 무선 백홀 트래픽을 전송하면 다른 EG 중 하나가 e₁과 e₆을 통해 통신을 시도 할 수 있다.In the scheme proposed according to an embodiment of the present invention, the ES and all EGs are equipped with a directional antenna. As shown in FIG. 2, ES exists with 5 EGs denoted by n ₀ to n _i , respectively. Set N = {n ₁ , n ₂ , ..., n ₅ } and

= {e ₁ , e ₂ , ..., e ₁₀ } indicates that the number of EGs (α) is 5 and the number of edges (β) is 10. The EG is equipped with a directional antenna consisting of four indexes. When n ₁ tries to send traffic to n ₅ , n ₂ and n ₄ cannot communicate because the target beam is blocked. However, if n ₁ transmits wireless backhaul traffic to n ₅ , one of the other EGs can try to communicate through e ₁ and e ₆ .

As a result, some EGs can transmit traffic simultaneously using a directional antenna, but all beams are not blocked. For example, e ₁ and e ₄ can communicate simultaneously without interference. Since the EG and its edges depend on the near-field environment, beam and channel scheduling is required for fair communication.

Crash / Hearing Loss Graph

In the scheme proposed according to an embodiment of the present invention, all EGs in the wireless backhaul are fixed, and their information is periodically exchanged with a hello message to establish a mesh network topology.

The hello message contains information such as EG position, data rate and beam index, and is carried over an omni-directional antenna. After receiving the latest hello message, EG updates with the information provided.

The ES can get the information of all EGs because it sends wireless backhaul traffic to the ES for all EGs to access the core network. Thus, the ES maintains a collision/hearing loss graph that determines whether different links are interfering with each other.

[Table 1]

As shown in Table 1, some edges cannot be scheduled at the same time. For example, if e ₆ connected between n ₃ and n ₄ is reserved, e ₁ or e ₄ can be reserved at the same time. However, since e ₂ interferes with other EGs except for e ₈ and e ₁₀ , it cannot be scheduled simultaneously with other edges.

This collision graph (C) is defined as follows.

[Equation 2]

Similar to the collision graph, the ES maintains a deafness graph, which determines if a hearing problem occurs when different links are scheduled at the same time. The hearing loss graph (F) is defined as follows.

[Equation 3]

Link scheduling algorithm based on transmission estimation

An embodiment of the present invention can be implemented as a link scheduling algorithm for formulating an objective function of a wireless backhaul network in the form of an MWIS problem and solving the MWIS problem.

Formulate the problem

In the algorithm according to an embodiment of the present invention, it is assumed that the ES has the information of all EGs and all the EGs communicate with each other under the control of the ES, so that the ES plays a role of scheduling a wireless backhaul network link. Firstly, the set of edges is represented by W = {w ₁ , w ₂ , ..., w _β }. Where β is the number of edges of the wireless backhaul. Then, the weight of the edge e in time slot t, w _e (t), is given as follows.

[Equation 4]

Here, Q _e (t) is the sum of the backlog size at the edge e, and r _e (t) represents the achievable rate of the link e in the time slot t defined by Equation 1. The definition of the weight is related to the instantaneous queue backlog size that changes over time. Thus, this definition differs from the definition of transmission throughput, which by definition is a time averaged amount. The queue back log size at the edge e evolves according to Equation 5 below.

[Equation 5]

　

Where μ _e (t) and λ _e (t) denote the number of bits leaving the queue of e and the number of bits added to the queue of e, respectively.

Further, it is assumed that all time slots t ∈ T occupying the channel have the same length, where T is a set of time slots. Conceptually, the purpose of the proposed scheduling algorithm is to find links, channels, data rates and routing flows that maximize the sum of weights for all possible independent sets. Therefore, the objective function is designed to find a link set that maximizes the weight as follows.

[Equation 6]

는 지표 변수이며 두 가지 경우로 표현 될 수 있다.here

Is an indicator variable and can be expressed in two cases.

[Equation 7]

However, in a smart manufacturing system, the problem of simply finding a link set that maximizes weight is practically impossible. Therefore, constraints such as hearing impairment management, collision management, and flow preservation may be considered in the present invention.

로 표시한다. 그런 다음, 다음의 흐름 보존 제약 조건은 다음과 같이 주어진다.1) Flow conservation constraint: Each flow f is transmitted at a specific data rate on link e during time slot t along multiple paths from source node s(f) to destination node d(f). Directional link from node n to node m

Denoted by Then, the following flow conservation constraint is given as

[Equation 8]

Where d _f represents the average data rate of flow f.

2) Link capacity restrictions: Since the average data rate through each directional link cannot exceed the average capacity through directional links, the following restrictions apply.

[Equation 9]

에 할당되는지 여부를 나타내는 채널 할당 표시 자로

를 정의한다. 즉, 아래 수학식 10과 같이 정의될 수 있다.3) Restriction of Channel Matching: Two nodes n and m must communicate through the same channel. Thus, channel k is a directional link

As a channel assignment indicator indicating whether it is assigned to

Defines That is, it may be defined as in Equation 10 below.

[Equation 10]

From Equation (10), we can derive Equation 11.

[Equation 11]

4) Channel restrictions: When different directional link pairs are allocated multiple times through the same channel, multiple channels can be allocated to the link. However, if the same channel is assigned to two or more pairs of directional antennas on a single link, interference occurs between transmissions from different pairs of directional antennas.

To solve this problem, different orthogonal channels must be assigned to a pair of directional antennas assigned to the same link. Therefore, the number of channels allocated to the directional link should be equal to the number of pairs of directional antennas allocated to the directional link.

[Equation 12]

에 할당되어야 한다.Here, M represents the number of antennas. In addition, when node n transmits data to node m through a directional link outgoing from channel k or receives data from node m through a directional link coming from channel k, link channel k.

Should be assigned to

Therefore, the following constraints are defined.

[Equation 13]

5) Transmission interference restriction: A node cannot simultaneously receive data through multiple receive directional links on the same channel. Therefore, the following feasible scheduling constraints are defined to avoid collision/hearing impairment between transmitting nodes and to ensure a feasible link scheduling scheme.

[Equation 14]

+

≥1,이라면 e'는 N (e)에 포함된다.Here, N(e) represents a set of neighboring edges for e.

+

If ≥1, e'is included in N (e).

6) Transmission Fairness Constraints: Each node consumes energy whenever it receives/transmits data. In a smart manufacturing environment, the amount of residual energy in each EG is important for efficient communication and transmission/reception of real-time data. Thus, fairness constraints confirm the opportunity for all EGs to be transmitted evenly and equalize energy consumption between EGs fairly. The following fairness constraints are set using Jain's Fairness Index.

[Equation 15]

Here, τ represents the fairness threshold.

7) Objective function: The objective function from Equation (8) to Equation (15) can be designed as follows.

[Equation 16]

Centralized link scheduling problem

The objective function (16) is complex and requires considerable time to solve using a variety of solutions.

Therefore, in order to reduce the scale of the problem, a single channel environment was assumed, and the amount of data transmitted in a time slot cannot exceed the channel capacity. Therefore, in the present invention, flow conservation, link capacity, and channel matching restrictions were removed, and the target function could be reduced as a pure MWIS problem.

In order to solve the MWIS problem, MWIS is a well-known nondeterministic polynomial time-hardness (NP-hard) problem, and thus various heuristics and approximation algorithms are proposed in an embodiment of the present invention.

In the scheme proposed according to an embodiment of the present invention, the scheduling link is calculated in real time by the ES and transmitted to all EGs through the control channel. In addition, each EG reports the queue backlog size to the ES at regular intervals via the control channel. In an embodiment of the present invention, an underlay control channel using some of the frequencies of each channel is assumed.

3 is a code of a link scheduling algorithm according to an embodiment of the present invention, FIG. 4 shows a plurality of nodes and links performing link scheduling according to an embodiment of the present invention, and FIGS. 5 and 6 are Represents a time slot representing link scheduling according to an embodiment of.

A sequence of link scheduling according to an embodiment of the present invention will be described with reference to FIGS. 3 to 6. First, the EG with the highest number of neighboring EGs is found (row 5). If more than one EG is found, the algorithm selects an EG with a small number of linked EGs. Otherwise, EG is chosen randomly (rows 6-11). The edge with the highest weight connected to the selected EG is scheduled first and all edges are assigned time slots in this way (lines 14-22). When all edges of the selected EG are scheduled, the corresponding EG is removed from the candidate list and the process continues until all edges of the EG are scheduled. In particular, in the process of scheduling other edges, the algorithm maximizes space reuse and finds a time slot to allocate it. When all edges are scheduled, we measure the fairness between EGs using Jain's fairness index. The scheduling step before the fairness measurement is performed is illustrated in FIG. 5, and the scheduling step in which fairness is measured and the fairness constraint is considered is illustrated in FIG. 6.

If the fairness constraint is less than a predetermined value, the edge with the lowest weight is found and scheduling continues (lines 24-27). Therefore, as shown in FIG. 6, the edge with the lowest weight is scheduled in the added time slot by adding a new time slot (t ₆ , t ₇ ), and if the fairness constraint is less than a predetermined value, the step is repeated. do.

The proposed heuristic algorithm efficiently handles complex MWIS problems and provides backhaul routing link scheduling results within a short time. It is suitable for intelligent manufacturing network systems to process commands and monitor the status of the ME in real time.

In order to evaluate the performance of the heuristic algorithm in terms of complexity, the computational complexity of each step was analyzed as follows. In the scheduling step, a target EG search was used as an example, and N comparisons to find the target EG are required. The target EG has the maximum number of adjacent nodes. However, in the worst case, N nodes may have the same number of neighboring nodes. Hence, the scheduling algorithm is quadratic for the size of N. Finally, the overall algorithm can take computational complexity as O(N ² ).

Performance evaluation

The proposed algorithm was implemented using the OPNET simulator. The simulation focused on the results of total network throughput, delay, packet loss rate and fairness. To perform the simulation, we assumed that a wireless backhaul network topology was built according to Figure 2. The detailed configuration of the simulation parameters is shown in Table 2.

[Table 2]

Also, in the simulation, we assumed that all EGs have already acquired routing information and set the scheduling delay to 10ms. Also, routing information is implemented by the simulator's matrix.

7 shows a graph of backhaul network throughput versus normalized traffic load.

For example, traffic load 2 means that incoming traffic is twice the average capacity of a single link on each EG in the backhaul system. Throughput of all the methods used increased rapidly as the traffic load increased. In particular, the indiscriminate substitution method showed the highest increase rate among all plans. In a situation where there is a heavy traffic load, the algorithm set to τ = 0.9 according to an embodiment of the present invention achieves more than 27% of the existing scheme.

However, when a large number of nodes are deployed, the brute force substitution method is not scalable. The DSR-based backhaul routing method showed the lowest performance due to the overhead required for routing path setup. A spatial division multiple access (SDMA) based resource management scheme for 5G optimizes link scheduling, but the total throughput is less than that proposed according to an embodiment of the present invention.

This is because the SDMA-based resource management scheme optimizes link scheduling considering only a single hop link. Thus, the frame can be lost in the intermediate GW and the end-to-end delay can increase.

Interestingly, it can be found that the performance of the algorithm proposed according to an embodiment of the present invention is better at higher τ. This means that high link-to-link transmission fairness increases end-to-end throughput.

8 shows the average delay versus normalized traffic load.

The average delay of the scheme proposed in FIG. 8 and the DSR-based scheme rapidly increases at traffic loads 2 and 0.5, respectively. We further investigated the delay performance at higher traffic loads and found that the delay of the proposed scheme (τ = 0.9) according to the present invention increases slowly when the traffic load is 3.5.

In a scheme with low fairness constraints (τ = 0.5 and τ = 0.7) according to another embodiment of the present invention, the average delay increases when traffic load = 2. This is due to unbalanced link scheduling that discards packets. In the DSR-based scheme, a node suffers signaling overhead because it has to find a routing path. As a result, network latency increases. In addition, since the radio resource management scheme for 5G performs link scheduling for one hop, a bottleneck occurs in the intermediate GW, indicating a high delay.

9 is a graph showing packet loss rate versus normalized traffic load.

The trend of the packet loss rate performance curve shown in FIG. 9 is similar to that of the delay performance. In the proposed scheme (τ = 0.9) according to an embodiment of the present invention, even when the traffic load is 4, packet loss hardly occurs. On the other hand, for the DSR-based scheme and the proposed scheme with low fairness constraints (τ = 0.5 and τ = 0.7), more than 63% and 30% of packets were discarded, respectively.

Table 3 shows Jain's fairness index when the traffic load is 5.

[Table 3]

As can be seen from the table, all schemes proposed according to an embodiment of the present invention almost satisfy the fairness constraint. The brute-force approach scores a top score of 0.9341. This result implies that the optimal value for τ in this scenario is 0.9341.

FIG. 10 shows backhaul network throughput versus normalized traffic load of a large network, and FIG. 11 shows average delay versus normalized traffic load in a large network.

10 and 11 respectively show throughput and delay results when 100GW is placed in a 1km×1km topology (a more stressful environment). As shown in the graph, the network throughput is similar to the previous result, and it can be seen that the throughput is saturated as traffic increases in all systems. In the case of delay, it can be seen that the overall increase. However, it can be seen that the delay of the scheme proposed according to the present invention is within 300 ms, which is the target end-to-end delay time proposed by EdgeX Foundry. Therefore, the technique proposed according to the present invention is suitable for real-time data processing in a smart manufacturing system.

The above description is merely illustrative of the technical idea of the present invention, and those of ordinary skill in the art to which the present invention pertains will be able to make various modifications and variations without departing from the essential characteristics of the present invention. Accordingly, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but to explain the technical idea, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be interpreted as being included in the scope of the present invention.

Claims (10)

In a link scheduling method for data processing in a backhaul network-based network system including a plurality of nodes,
Selecting any one node to perform scheduling among the plurality of nodes (a);
(B) sequentially scheduling a plurality of links connected to the selected node in consideration of collision-hearing loss constraints;
(C) repeating the scheduling step (b) from a first time interval by selecting any one of the remaining unselected nodes among the plurality of nodes;
(D) searching for a link that satisfies the collision-hearing-loss constraint for each time interval when scheduling of all links is completed once at a time;
(E) performing channel utilization by scheduling the searched link in a corresponding time interval
Calculating channel utilization for all time intervals (f);
Calculating a fairness constraint based on the channel utilization (g);
(H) selecting a link with the lowest channel utilization among the plurality of links when the fairness constraint is less than a predetermined value;
(I) scheduling the link in an additional time interval;
(J) repeating steps (f) to (h); And
And (k) terminating scheduling when the fairness constraint is greater than or equal to a predetermined value. The method of claim 1,
In the step of selecting any one node,
The link scheduling method, characterized in that the node selected as the node to perform the scheduling is a node having the most links. The method of claim 1,
In the step of selecting any one node, if all the nodes have the same number of links, a link scheduling method in which one node is randomly selected. delete delete delete The method of claim 1,
The step (i) of scheduling the link in an additional time interval
And (l) searching for and scheduling a link that satisfies the collision-hearing loss constraint for the additional time interval. The method of claim 1,
The network system is a link scheduling method, characterized in that the smart factory. The method of claim 1,
The link scheduling method in which the backhaul network uses a millimeter wave (mmWave) band. In the link scheduling apparatus for data processing in a network system based on a backhaul network including a plurality of nodes,
Including a first scheduling unit, a second scheduling unit and a third scheduling unit,
The first scheduling unit,
Selecting any one node to perform scheduling among the plurality of nodes,
The plurality of links connected to the selected node are sequentially scheduled in consideration of collision-hearing loss constraints,
Selecting any one of the remaining unselected nodes among the plurality of nodes and repeating the scheduling from the first time interval,
The second scheduling unit,
When the scheduling of all links is completed once, a link that satisfies the collision-hearing loss constraint is searched for each time interval,
Channel utilization is performed by scheduling the searched link in a corresponding time period,
The third scheduling unit,
Channel utilization is calculated for all time periods,
Calculate fairness constraints based on the channel utilization,
When the fairness constraint is less than a predetermined value, a link having the lowest channel utilization among the plurality of links is selected,
Scheduling the link in an additional time period,
Repeating the calculation of the channel utilization and the fairness constraint and selecting the link with the lowest channel utilization,
When the fairness constraint is greater than or equal to a predetermined value, a link scheduling apparatus for processing data to terminate scheduling.

Images