KR20160130593A - Packet Drop Method for Differentiated Services in Wireless/Wired Ship and Indoor Area Networks - Google Patents

Abstract

Provided is a packet discarding method for supporting a differentiated service in a ship and an indoor wired/wireless network. The method comprises the following steps of: separating flows, which belong to one service class, into a plurality of flow groups; and applying a different packet discharging probability (P_drop_max) to each separated flow group.

Description

TECHNICAL FIELD [0001] The present invention relates to a Packet Drop Method for differentiated services in a ship and an indoor wired /

The present invention relates to a packet discarding technique for supporting a differentiated service in a ship and an indoor wired / wireless network.

The ship's standard network is defined by IEC (International Electronics Committee) as IEC 61162-1, 2, 3, and 4.

IEC 61162-1 is NMEA 0183, which is a single talker, single listener serial data communication of 4,800 bps, which was mainly installed in ships' navigation equipment since the 1980s.

IEC 61162-2 is a high speed version of NMEA 0183, a 384 kbps single talker, a multi listener, and an electrical signal RS422.

NMEA 2000 is multi-talk, multi-listen, multi-master, PnP, the physical layer uses CAN 2.0B and standardized to IEC 61162-3. In addition to single packet and multi-packet of SAE J 1939, NMEA 2000 added a fast packet protocol to modify the instrument level network to be controlled in real time on the ship.

In addition, IEC 61162-4 is an ethernet-based Maritime Information Technology Standards (MiTS) protocol initiated in Norway and is used as a standard for the shipboard control network.

The Internet Engineering Task Force (IETF) proposed Integrated Service (IntServ) and Differentiated Service (DiffServ) as the next generation network structure to guarantee QoS required by users.

IntServ guarantees end-to-end QoS for individual flows but all routers passing through the flow must maintain state information for each flow and Resource ReSerVation Protocol (RSVP) as a method for admission control. There is a problem in scalability. IntServ also needs to modify the existing infrastructure of the Internet.

The DiffServ scheme uses per-class management for each set of classes rather than management for each user flow, and defines Per Hop Behavior (PHB) for IP packets using DiffServ Code Point (DSCP).

All packets with the same DSCP code are processed in the same way. This method is an aggregation mechanism that classifies and processes a large number of user flows into a small number of classes and has scalability suitable for an internal network that delivers a large number of flows.

User flows are shown as In-Profile (In) and Out-of-Profile (Out) packets according to the contracted Traffic Profile compliance at the entrance router of the DiffServ domain, and RED In and Out (RIO) And protects the In packet by discarding the Out packet preferentially. In the wired and wireless network onboard, it functions to autonomously manage various sensors and controllers on board and to provide remote control.

If differentiated services are supported in such wired / wireless communication networks, high-priority real-time flows can be guaranteed to be transmitted at a minimum rate, and thus, an intelligent ship communication network can be realized.

SUMMARY OF THE INVENTION Accordingly, it is an object of the present invention to provide a packet disposal method for supporting a differentiated service in a ship and an indoor wired / wireless network that provides packet discarding technology superior to existing RIO (RED In and Out) .

According to an aspect of the present invention, there is provided a method of discarding packets for supporting differentiation services in a ship and an indoor wired / wireless network, the method comprising: dividing flows belonging to one service class into a plurality of flow groups; And applying different packet discard probability (P_drop_max) for each flow group.

According to the present invention, in order to realize an intelligent ship communication network, a differential service is supported in a wired / wireless communication network in a ship so that it is divided into several flow groups according to the FEC standard without being managed for each flow, Thereby improving the yield rate of the liver.

In addition, it can be applied not only to Layer 3 network layer but also to Layer 2 link layer with respect to DiffServ communication technology.

In addition, it is possible to improve flow quality QoS performance in DiffServ that provides only CoS (Class of Service) of group unit.

In addition, it can be effectively applied to a wired / wireless network in a ship in which a flow of small and medium units flows like a simulation environment.

FIG. 1 is a diagram showing an existing in-ship wired / wireless network (MiTS) and NMEA network configuration.
FIG. 2 is a graph showing a yield measurement result according to the number of bottleneck sections (** yield measurement result in a network over-provisioned at 160%).
3 is an algorithm for calculating P_drop_max according to an embodiment of the present invention.
4 is a graph showing the relationship between drop_index and P_drop_max.

BRIEF DESCRIPTION OF THE DRAWINGS The above and other objects, advantages and features of the present invention and methods of achieving them will be apparent from the following detailed description of embodiments thereof taken in conjunction with the accompanying drawings.

The present invention may, however, be embodied in many different forms and should not be construed as limited to the exemplary embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete and will fully convey the concept of the invention to those skilled in the art. And effects of the present invention, the scope of the present invention is defined by the description of the claims.

It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. In the present specification, the singular form includes plural forms unless otherwise specified in the specification. &Quot; comprises "and / or" comprising ", as used herein, unless the recited component, step, operation, and / Or added.

The packet discarding technique according to the present invention is not managed for each flow, but is divided into several flow groups according to specific criteria, and provides fairness between flows in the same class through group management.

The present invention reduces the link utilization rate slightly more than the RIO scheme, but the flows passing through the multiple bottleneck intervals can be guaranteed to have a higher data rate than the RIO scheme.

Also, the DiffServ communication technology according to the present invention can be applied not only to the layer 3 network layer but also to the layer 2 link layer.

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

FIG. 1 is a diagram showing an existing in-ship wired / wireless network (MiTS) and NMEA network configuration.

1, the existing wired and wireless wired and wireless network and the NMEA network are composed of an NMEA device network in which a gateway is connected to an in-line control network (Ethernet-based MiTS) and each device is connected to the in- .

According to an embodiment of the present invention, intelligent intra-ship communication network is provided by providing differentiated QoS to various intra-line communication traffic flowing into each gateway. That is, in the gateway of FIG. 1, the RIO and the packet discarding technique according to an embodiment of the present invention are applied to guarantee a data rate differentiated in units of a flow group.

Even within one domain, the paths of each flow are so diverse that the RTT (Round Trip Time) between each flow and the bottleneck interval experienced during arrival to the destination are different.

As can be seen from the simulation result of FIG. 2, it can be seen that in the case of a flow having a long RTT and a large number of hops, the yield is relatively decreased as compared with the flow which is not. The effect is more certain when there are many hops in the bottleneck.

It is assumed that the network applied to an embodiment of the present invention is appropriately bandwidth provisioning.

In the existing DiffServ network, if all the flows have sufficient bandwidth, they can satisfy the contracted target rate. However, if the bandwidth is sufficient, a flow having a small number of bottleneck segments occupies most of the surplus bandwidth, and a flow having a large number of bottleneck segments will only occupy a surplus bandwidth so as to match the target data rate.

As can be seen from FIG. 2, in the absence of a packet dropper, a 60% surplus bandwidth is required for all flows to satisfy the target transmission rate. However, in this case, it can be seen that the flow passing through one bottleneck section has a yield of 2.3 times the target transmission rate.

Because of this problem, even if a contract is made for the same service class, there is a difference in the QoS that the actual flow will receive depending on the situation experienced during arrival to the destination. This problem is caused by providing only Class of QoS (CoS), which is a basic purpose of DiffServ.

The Intelligent Drop Algorithm proposed in the present invention performs control on a flow group-by-flow basis without performing flow-by-flow management. We want to divide (divide) the flows belonging to one service class into several flow groups (or a plurality of flow groups) according to a certain criterion, thereby reducing inequity among the groups so as to provide yield fairness between flows.

According to the present invention, priorities are assigned according to the traffic characteristics occurring in each Least Terminal as a criterion for dividing a flow group, and a label is attached thereto, and a network layer or a link, such as Multi-Protocol Label Switching (MPLS) And performs switching in the layer. This approach is not affected by the number of flows, so there is no scalability problem.

The key point of the Intelligent Drop Algorithm is that packet dropping probability is different for each of these flow groups. That is, we classify flow groups according to the labels generated at each end terminal and apply different P_drop_leaf (leaf early drop probability) to each group to solve the unfairness among the flows flowing into the network domain.

Each end terminal allocates a label according to the priority of the traffic to be transmitted, marks it as an In-profile packet by the amount of the target transmission rate to be contracted, and marks the exceeded packet as Out (Out-of-profile) do.

In addition, packet discarding is performed in advance so as not to have a data transmission rate which is much higher than the contracted transmission rate. If only the unintended target transmission rate is used as a criterion and the packets of the flow group exceeding the target transmission rate are discarded, the network resources may not be used sufficiently. However, in the present invention, packet disposal is scarcely generated if the remaining bandwidth is equally divided . Therefore, there is no big problem in terms of the link utilization rate.

The gateway applies a different P_drop_max (max drop probability) for each input interface classified according to the type of the label, thereby resolving the unfairness among the flows passing through the gateway.

In the conventional RIO buffer applied to the gateway, packet discarding is performed according to Equation (1) below. In the present invention, however, P_max_drop is used for each input interface so that the fair bandwidth of each interface is used, I want to.

The calculation method of P_drop_leaf and P_max_drop are the same, and there is only a difference in the method of application, and the pseudo code of the algorithm for calculating P_drop_max is shown in FIG.

3, group_n. share_index is the number of Out packets per In packet among the packets transmitted by group_n flows.

The above equation (1) is the most basic equation for obtaining the packet discard probability. share_index has several advantages because it takes the form of a ratio. In other words, there must be a clear difference in the amount of excess bandwidth between the contracted flow for 1Mbps traffic and the contracted traffic for 2Mbps. Of course, the bandwidth occupancy ratio should be about twice. If the flow of 1Mbps traffic occupies a surplus bandwidth of 0.5Mbps, the flow of 2Mbps traffic occupies a surplus bandwidth of 1Mbps to have the same share_index value.

group_n. share_ratio is a value indirectly representing the amount of current occupied by the group_n in total surplus bandwidth.

If there are 10 groups in total (N = 10) and 10 groups all occupy the same amount, the share_ratio of all groups will be 0.1. Also, if it occupies a large amount, it will be larger than 0.1, and in the opposite case, it will be smaller than 0.1. If so, the value of drop_index will be greater than 1 if occupied by a large number of expressions, and less than 1 if it is otherwise. This value becomes the input value of the function that determines P_drop according to the following formula.

4, the difference between the values of drop_index greater than 1.0 and smaller values of P_drop_max is large.

If drop_index is 1, it corresponds to a group that has fairly distributed the remainder bandwidth as described above. In this case, P_balance is packet discard probability.

However, if drop_index is larger than 1, that is, occupies a large amount of surplus bandwidth, the probability of packet dropping exponentially increases exponentially.

In the opposite case, the packet drop probability decreases exponentially. Therefore, if a large amount of surplus bandwidth is occupied according to such rules for each group, the probability of packet discarding is high, and the yield is lowered, and other flows can be taken by the other flows.

In Leaf Terminal, packets are not discarded for In packet because it plays a role to remove unfair part in advance before packet discard by congestion occurs in internal gateway.

However, based on the above P_drop_max, the packet disposal for Out packet is performed. However, as described above, since the packet is discarded before congestion occurs, if the probability of discarding a packet is too large, the overall yield is lowered. Therefore, it is necessary to find an appropriate upper limit value. Therefore, P_drop_max is multiplied by 0.1, which is the coefficient that can be adjusted.

The present invention is applied to a case where a 2.5-layer protocol supporting several fixed paths such as MPLS supports a 3-layer protocol.

In order to define the label, the most commonly used method is to define the same flow as the <Ingress terminal, Egress terminal> pair as a FEC (Forwarding Equivalence Class) and assign the Label assigned to the destination address.

When the present invention is applied in such an environment, the traffic is controlled in a direction to divide the surplus bandwidth occupied by the groups in units of FECs uniformly. This greatly reduces the complexity when compared with the case of controlling by flow unit. In the gateway, the present invention has a characteristic of congestion control. Currently, the congestion control technology that is most widely used is RIO. In the present invention, a part of RIO technology is applied.

Since it is the main purpose of the gateway to minimize the congestion situation while minimizing packet discard, it is necessary to adjust the packet discard probability in consideration of the situation of the RIO queue. Therefore, we use the P_drop equation used in RIO in Equation (1). Unlike the existing RIO technology, the value of P_max_drop depends on the input interface.

While the present invention has been described with reference to exemplary embodiments, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, .

Claims (2)

Dividing flows belonging to one service class into a plurality of flow groups; And
Applying different packet discard probability (P_drop_max) for each divided flow group;
A method for discarding packets for supporting differentiated services in a ship and an indoor wired / wireless network.
2. The method of claim 1, wherein dividing into the plurality of flow groups comprises:
Assigning priorities according to traffic characteristics occurring in Leaf Terminals;
Generating a label according to the priority; And
Dividing the flows into a plurality of flow groups according to the generated label
The method of claim 1, wherein the packet discarding method is for supporting differentiated services in a ship and an indoor wired / wireless network.

Images