WO2020233114A1 - 车联网下安全型防御合谋攻击的系统及其方法 - Google Patents

车联网下安全型防御合谋攻击的系统及其方法 Download PDF

Info

Publication number
WO2020233114A1
WO2020233114A1 PCT/CN2019/127707 CN2019127707W WO2020233114A1 WO 2020233114 A1 WO2020233114 A1 WO 2020233114A1 CN 2019127707 W CN2019127707 W CN 2019127707W WO 2020233114 A1 WO2020233114 A1 WO 2020233114A1
Authority
WO
WIPO (PCT)
Prior art keywords
ioc
reputation
providers
vehicles
internet
Prior art date
Application number
PCT/CN2019/127707
Other languages
English (en)
French (fr)
Inventor
赵锋
冯景瑜
Original Assignee
西安安盟智能科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 西安安盟智能科技股份有限公司 filed Critical 西安安盟智能科技股份有限公司
Priority to US17/612,720 priority Critical patent/US12010518B2/en
Priority to JP2022516253A priority patent/JP7407913B2/ja
Publication of WO2020233114A1 publication Critical patent/WO2020233114A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W12/00—Security arrangements; Authentication; Protecting privacy or anonymity
    • H04W12/12—Detection or prevention of fraud
    • H04W12/121—Wireless intrusion detection systems [WIDS]; Wireless intrusion prevention systems [WIPS]
    • H04W12/122—Counter-measures against attacks; Protection against rogue devices
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/40—Services specially adapted for particular environments, situations or purposes for vehicles, e.g. vehicle-to-pedestrians [V2P]
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60R—VEHICLES, VEHICLE FITTINGS, OR VEHICLE PARTS, NOT OTHERWISE PROVIDED FOR
    • B60R25/00—Fittings or systems for preventing or indicating unauthorised use or theft of vehicles
    • B60R25/30—Detection related to theft or to other events relevant to anti-theft systems

Definitions

  • the present invention relates to the field of vehicle networking technology, and the present invention also relates to the field of defense technology, in particular to a system and method for secure defense against collusion attacks under the vehicle networking.
  • the Internet of Vehicles is a huge interactive network composed of vehicle location, speed, and route information.
  • the vehicle Through GPS, RFID, sensors, cameras, image processing devices, etc., the vehicle can complete the collection of its own environment and status information.
  • Each vehicle can transmit these information to the central processing unit.
  • the central processing unit Through the analysis and processing of the central processing unit, it can calculate the best route for different vehicles and report road conditions in time, which is beneficial to the car networking system to arrange the signal light cycle, etc. .
  • V2V Vehicle-to-vehicle
  • V2I vehicle-to-infrastructure
  • DSRC Dedicated short-range communication
  • Vehicles are equipped with on-board units, namely sensors, resource command processors, storage and communication equipment for data collection, processing and sharing. With the help of the on-board unit, the vehicle can automatically detect traffic-related events and use vehicle-to-vehicle communication standards to send warning messages to others. This information helps the vehicle to understand the traffic situation in time, thereby improving the efficiency of traffic safety.
  • on-board unit namely sensors, resource command processors, storage and communication equipment for data collection, processing and sharing.
  • the vehicle can automatically detect traffic-related events and use vehicle-to-vehicle communication standards to send warning messages to others. This information helps the vehicle to understand the traffic situation in time, thereby improving the efficiency of traffic safety.
  • each car can play two roles, namely, the role of providing messages and the role of receiving messages.
  • V a the vehicle (V a ) broadcasts an out-of-control warning message to warn the vehicles behind it.
  • V b receives this warning message, it is vital for V b to judge the credibility of this message and make a quick decision. Due to time constraints, it is impractical to verify the authenticity of the information from neighboring vehicles or trusted third parties. If this warning message is false, then the V b brake will be dangerous. Therefore, in the Internet of Vehicles, how to effectively establish the trust relationship between vehicles is very important. We hope that every vehicle can detect malicious vehicles and their reported false information.
  • the reputation system (also known as the trust management scheme) enables the vehicle to determine whether the received message is credible, and provides a basis for the network operator to reward and penalize a specific vehicle.
  • the reputation score of a vehicle can be calculated by scoring its past behavior.
  • TMAM traffic-related message aggregation model
  • consumers using the reputation system can collect all information about a certain traffic incident, filter out false information reported by malicious vehicles, and only detect credible information. Then use a specific model to aggregate all traffic messages to make quick decisions, such as the principle of majority decision.
  • the traffic-related message aggregation model (TMAM) can be managed by the cloud server of the central car network, or it can be managed by the RSU in the distributed car network.
  • consumers can generate ratings for the message according to the final result of the event, and upload these ratings to update the reputation score of the vehicle providing the message.
  • IOC inside-and-outside Collusive Attack
  • R i > ⁇ indicates a higher reputation score, which means that the vehicle's message can be accepted by TMAM. This allows IOC attackers to find an attack process to prompt their reputation score.
  • R i ⁇ + ⁇ 1 V i stop fake traffic-related information, and asked one of his side's accomplice IOC to launch attacks, to enhance his reputation score.
  • reputation systems can be divided into two categories: centralized reputation systems and decentralized reputation systems.
  • a centralized reputation system In a centralized reputation system, all ratings are stored and processed in a central server, such as a cloud server. Since vehicles usually need to make decisions within a relatively short delay, these centralized systems are not always able to meet the stringent quality of service (QoS) requirements of the Internet of Vehicles.
  • QoS quality of service
  • the reputation calculation task is performed in the vehicle itself or in the RSU. Local management of reputation scores may reduce interaction with network infrastructure.
  • the reputation system one of the most popular designs is based on beta functions. It first calculates the number of credible and malicious behaviors performed by the vehicle. Then use the beta function to calculate the reputation score.
  • V i When a message (m i) is reported as V i, if the authenticity of the traffic-related information m i and consumer evaluation of the same, the m i is credible information. Otherwise, it will be considered false information.
  • the vehicle V i the credibility of the system calculates the number of trusted message (cre i) the report of the vehicle, and the vehicle information of the number of false report inc i.
  • the reputation score of Vi is calculated using the beta function as:
  • the reputation system can be used by IOC attackers to improve their reputation scores and thus manipulate TMAM.
  • the present invention provides a system and method for secure defense against collusion attacks under the Internet of Vehicles, which effectively avoids that the reputation system in the prior art can be used by IOC attackers to improve their reputation scores and manipulate TMAM. defect.
  • the present invention provides a security system and method for defending against collusion attacks under the Internet of Vehicles, specifically as follows:
  • a security system for defense against collusion attacks in the Internet of Vehicles including the cloud server of the Internet of Vehicles platform or the RSU in the distributed Internet of Vehicles;
  • the system for security-type defense against collusion attacks in the Internet of Vehicles also includes an on-board unit installed on the vehicle, the on-board unit is connected to the cloud server of the Internet of Vehicles platform or the RSU in the distributed Internet of Vehicles, and the on-board unit includes processing
  • the processor is connected with the wireless communication module, and the processor is also connected with the memory.
  • the cloud server of the Internet of Vehicles platform or the RSU in the distributed Internet of Vehicles includes a traffic-related message aggregation model TMAM;
  • the related message aggregation model TMAM is used to store historical messages from providers and historical rating data of consumers at the end of a traffic-consuming voting campaign.
  • the method for the system of security type defense against collusion attacks under the Internet of Vehicles runs on the cloud server of the Internet of Vehicles platform or the RSU in the distributed Internet of Vehicles, and the method includes the following methods:
  • the RFAA method of reputation fluctuation correlation analysis is used to detect IOC attacks, analyze the characteristics of reputation fluctuations, and delete unlikely IOC providers from all providers in the current voting operation of events that consume traffic.
  • the reputation fluctuation correlation analysis RFAA includes the following methods:
  • the data management supporting detecting IOC attacks includes a two-step processing message report and consumer rating.
  • the two-step processing message report and consumer rating method is specifically as follows:
  • the traffic-related message aggregation model TMAM should store historical messages from providers and historical rating data of consumers at the end of the traffic-consuming voting activities.
  • the providers are the on-board units of vehicles for providing messages.
  • the on-board unit of the vehicle is used to receive and rate the messages, so as to realize the message report based on two-step processing;
  • Each vehicle V i is assigned to a table PC, the PC table holds provider previous news reports on the vehicle, and when the vehicle acts as a consumer, all providers of news and rating of V i All are stored in the PC table, thus achieving consumer ratings;
  • the P-C table is shown in Table 1:
  • Vt represents the polling hours when the vehicle V i request voting behavior and traffic related events, vt i h is the first vehicle V i h times in the latest voting time, h is a natural number;
  • P_ID (message) represents the ID assigned to the provider and its historical information.
  • V j (m j ) h
  • V j is the ID of the j-th provider
  • (m j ) h is the ID of the For messages in traffic-related event voting operations
  • i, j, and n are natural numbers and are serial numbers assigned to the vehicle;
  • C_ID (rating data) indicates the ID of consumers rating their history, for V i (t) h is, V i is the i-th consumer's ID, (t) h h for the first time to vote in traffic-related incidents
  • the corresponding real level in the behavior, i, j, and n are natural numbers and are the serial numbers assigned to the vehicle.
  • the method for obtaining association rules of reputation fluctuation includes:
  • Step 1-1 Initialize the value of the index variable ⁇ i as the reputation fluctuation index to 0;
  • Step 1-2 Set the initial value of the integer variable k to 1;
  • the RFAA method of reputation fluctuation association analysis further includes analyzing the association relationship between IOC attackers, and the method of analyzing the association relationship between IOC attackers is specifically as follows:
  • the index support number s( ⁇ ) can be used to identify frequent providers
  • IOC consumers Describe the relationship between IOC consumers and IOC providers. It can also be found that IOC users and IOC providers often appear at the same time, and IOC consumers that often appear at the same time as IOC providers can be identified;
  • the IOC attackers can successfully increase their reputation scores.
  • V i suppose is a consumer, and ⁇ is a group of providers and the current consumption of traffic incidents voting operations,
  • the reputation fluctuation association rule can be designed as shown in formula (5):
  • the formula (5) is constrained at ⁇ j ⁇ 2, wherein, ⁇ j is the index variable V j reputation provider fluctuation index, V i for the consumer, the provider V j, the minimum support count minsup ; Providers that meet the above formula (5) are classified as IOC attackers.
  • the method for analyzing the association relationship between IOC attackers further includes a dynamic sampling observation method, and the dynamic sampling observation method is:
  • Step 2-1 The initial minsup value is set to a preset value
  • Step 2-2 View the number of detected IOC attacks h;
  • Step 2-3 Use The function output supports the average value of the count and use it
  • the function gets the minimum support count, uses the minimum support count to update the value of minsup, and then goes to step 2-5 to execute;
  • Step 2-4 Also use To judge whether the judgment formula is established, if not, perform steps 2-5, if the judgment formula is established, use The function gets the minimum support count, uses the minimum support count to update the value of minsup, and then goes to step 2-5 to execute;
  • Step 2-5 View the other h detected IOC attacks, use The function gets the average value of the support count, and then returns to step 2-4 for execution, where l, h, and m are natural numbers.
  • the method for deleting the unlikely IOC provider includes the following steps:
  • Step 3-2 Traverse each provider V j that belongs to ⁇ , and judge one by one whether its ⁇ j ⁇ 2, if yes, then ⁇ 1 ⁇ V j ⁇ belongs to ⁇ , where ⁇ j is the reputation of provider V j Index variables of the volatility index;
  • the present invention can repair the loopholes in the reputation and reputation system in the Internet of Vehicles, and the IOC attacker can manipulate the TMAM by enhancing his own reputation score through internal and external conspiracy. It can quickly detect IOC attacks and improve the security of the Internet of Vehicles. By recursively eliminating the suspicious provider and the proposed association rules for reputation fluctuations, the overload of TMAM is avoided. It can deprive IOC attackers of the opportunity to prompt their reputation scores and ensure credible information in the Internet of Vehicles. Without interference from IOC attackers, the fairness and availability of TMAM are guaranteed.
  • Figure 1 is a schematic diagram of a car networking system model.
  • Figure 2 is a schematic diagram of the IOC attack strategy.
  • FIG. 3 is a schematic diagram of the structure of the reputation fluctuation correlation analysis RFAA of the present invention.
  • Fig. 4 is a working flow chart of the dynamic sampling observation method of the present invention.
  • FIG. 5 is a flow chart of the IOC attack detection of the present invention.
  • the security-type defense against collusion attacks under the Internet of Vehicles can cause two-dimensional damage to the performance of TMAM, and IOC attackers are more likely to damage the fairness and availability of TMAM.
  • IOC consumers can send ratings based on information from co-conspirators. After several rounds of attacks, IOC attackers can help each other to quickly obtain a higher reputation score, thereby making it easier to operate TMAM. In the prior art, the attacker only plays the role of the provider, sometimes relying entirely on his honest behavior to improve his reputation score.
  • IOC attacks may affect the sentiment of trusted vehicles because they may not use trusted information to improve their reputation score. If their credible information is different from the ratings of IOC consumers, their reputation scores will not be improved. In the prior art, credible information can improve the credibility score of credible vehicles.
  • the core of the security collusion attack defense system and method under the Internet of Vehicles of the present invention is to use the design idea of reputation fluctuation association analysis (RFAA) to design a defense scheme for quickly detecting IOC attacks.
  • RFAA reputation fluctuation association analysis
  • Reputation fluctuation correlation analysis RFAA is used to detect IOC attacks, analyze the characteristics of reputation fluctuations, and delete unlikely IOC providers from all providers in the current and traffic-consuming event voting operations.
  • the quantitative method of credit fluctuation characteristics is given in the RFAA program of credit fluctuation correlation analysis.
  • the minority cannot change the decision of the traffic-related message aggregation model TMAM according to the principle of majority. If the number of credit fluctuation characteristics of a provider is smaller than that of most providers, the current voting behavior of traffic-consuming events is safe and not subject to IOC attacks, because it is impossible for a few people to use the traffic-related message aggregation model TMAM.
  • cluster analysis is used to detect collusion attacks, and it is impossible to discover in advance whether a suspicious provider is a minority of all providers. Therefore, the prior art will waste more time to detect whether the current voting behavior of traffic-related events is safe.
  • the search volume of the database can be reduced, and the reputation fluctuation association rules for detecting IOC attacks proposed in the reputation fluctuation correlation analysis RFAA scheme can be supported.
  • the present invention detects an IOC attack, it should abandon the current traffic-related event voting behavior, thereby depriving the IOC attacker of the opportunity to improve his reputation score.
  • it is necessary to detect all colluding attackers, which may also increase the overload of the traffic-related message aggregation model TMAM.
  • a security system for defense against collusion attacks in the Internet of Vehicles including the cloud server of the Internet of Vehicles platform or the RSU in the distributed Internet of Vehicles;
  • the system for security-type defense against collusion attacks in the Internet of Vehicles also includes an on-board unit installed on the vehicle, the on-board unit is connected to the cloud server of the Internet of Vehicles platform or the RSU in the distributed Internet of Vehicles, and the on-board unit includes processing
  • the processor can be a PLC, a single-chip microcomputer, a DSP processor or an ARM processor, the processor is connected to a wireless communication module, and the wireless communication module can be a 3G module or a 4G module connected to an Internet of Vehicles platform.
  • the processor is also connected to a memory, and the memory can be a flash memory or an external memory.
  • the cloud server of the Internet of Vehicles platform or the RSU in the distributed Internet of Vehicles includes a traffic-related message aggregation model TMAM;
  • the related message aggregation model TMAM is used to store historical messages from providers and historical rating data of consumers at the end of a traffic-consuming voting campaign.
  • the method for a system for security-type defense against collusion attacks under the Internet of Vehicles runs on the cloud server of the Internet of Vehicles platform or the RSU in the distributed Internet of Vehicles, and the method for the system for security-type defense against collusion attacks under the Internet of Vehicles includes As follows:
  • the RFAA method of reputation fluctuation correlation analysis is used to detect IOC attacks, analyze the characteristics of reputation fluctuations, and delete unlikely IOC providers from all providers in the current voting operation of events that consume traffic.
  • the voting operation of the traffic-consuming event mainly refers to the road condition interaction event, that is, the voting behavior of traffic-related events.
  • the reputation fluctuation correlation analysis RFAA includes the following methods:
  • the data management supporting detecting IOC attacks includes a two-step processing message report and consumer rating.
  • the two-step processing message report and consumer rating method is specifically as follows:
  • the traffic-related message aggregation model TMAM should store historical messages from providers and historical rating data of consumers at the end of the traffic-consuming voting operation, instead of discarding them.
  • the provider is the on-board unit of the vehicle for providing Message
  • the consumer is the on-board unit of the vehicle for receiving messages and rating the messages, thus realizing a two-step processing message report;
  • each vehicle V i can be assigned to a table PC, the PC table holds news provider reported previously on the vehicle, when the vehicle acts as a consumer, to take the vehicle V i
  • all providers of news and ratings of V i are stored in the PC table, thus achieving consumer ratings;
  • the obtained association rule of reputation fluctuation supports the detection of IOC attack, and the association rule of reputation fluctuation is determined by analyzing the characteristics of the IOC attack.
  • the P-C table is shown in Table 1:
  • sn represents the voting behavior of the number of traffic-related event, the number of votes actions related to traffic incidents vehicle V i h is initiated;
  • vt represents the polling hours when the vehicle V i request voting behavior and traffic related events, vt i h is the first vehicle V i h times in the latest voting time, h is a natural number;
  • P_ID (message) represents the ID assigned to the provider and its historical information.
  • V j (m j ) h V j is the ID of the j-th provider, and (m j ) h is the ID of the traffic-related event message in the voting operation; specifically, when V i reports "1", V j (m j) h is recorded as V j (1) h, V i when report "0", V j (m j) h is recorded as V j (0) h, when no report V i, V j (m j) h is recorded as V j (-) h;
  • C_ID (rating data) indicates the ID of consumers rating their history, for V i (t) h is, V i is the i-th consumer's ID, (t) h h for the first time to vote in traffic-related incidents
  • the corresponding real level in the behavior, i, j, and n are natural numbers and are the serial numbers assigned to the vehicle.
  • the method for obtaining association rules of reputation fluctuation includes:
  • IOC attackers will report credible messages to maintain a high reputation score, or report credible messages to manipulate TMAM.
  • IOC attackers have the characteristics of reputation fluctuations, so the indicator variable ⁇ of the reputation fluctuation index is introduced to quantify the characteristics of IOC attackers’ reputation fluctuations.
  • the index ⁇ can be calculated by observing whether the reputation score of a car continuously increases or decreases. .
  • V i, R k i represents the vehicle V i k th reputation score.
  • Step 1-1 the variables ⁇ i as an index value of credit volatility index is initialized to 0, where, ⁇ i represents the index of the fluctuation of the vehicle V i credibility;
  • Step 1-2 Set the value of the integer variable k to 1;
  • the updating ⁇ i can be performed at the end of each traffic-related event voting action, in order to avoid the redundancy of calculating ⁇ i when detecting IOC attacks.
  • ⁇ i ⁇ 2 this means that an attacker has launched at least one round of IOC attacks to improve his reputation score and reported false information.
  • some malicious vehicles that behave honestly can sometimes produce ⁇ i ⁇ 2. Therefore, based on the obtained reputation fluctuation association rules, the screening range of suspicious providers can be narrowed. In order to effectively detect IOC attackers, it is also necessary to analyze the relationship between IOC attackers.
  • the present invention analyzes the association relationship between IOC attackers based on the following three characteristics.
  • IOC attacker can play the role of IOC consumer and help his accomplices play the role of IOC provider. After a round of IOC attacks, one of the IOC providers will be designated as the new IOC consumer, and the former IOC consumer can become the new IOC provider.
  • TMAM's decision should be in agreement with the majority of people.
  • the RFAA method of reputation fluctuation association analysis further includes analyzing the association relationship between IOC attackers, and the method of analyzing the association relationship between IOC attackers is specifically as follows:
  • IOC providers can also be frequent providers that often appear together.
  • the index support number s( ⁇ ) can be used to identify frequent providers; for example, if V 1 , V 2 , and V 3 are three frequent providers. Their support count is s (V 1 , V 2 , V 3 ), which is the number of times that V 1 , V 2 , and V 3 appear at the same time. You should search the PC Table to satisfy the following rules
  • minsup is the minimum supported count.
  • V 6 is an IOC consumer and V 1 , V 2 ,V 3 are three IOC providers, and the relationship between them can be described as V 6 ⁇ V 1 ,V 2 ,V 3 ⁇ .
  • V 6 IOC if the number of IOC provider exceeds the current events related to voting transportation operations provide more than half of those who, IOC attacker can successfully enhance their credit score, such as V 6 IOC also consumers, and V 1 , V 2 , V 3 , V 4 , and V 5 are providers of voting activities for current traffic-related events. V 1 , V 2 , and V 3 can manipulate TMAM because they are most providers. According to the association analysis rule formula (3), V 6 and V 1 , V 2 , and V 3 can be identified as IOC attackers
  • the reputation fluctuation association rule can be designed as shown in formula (5):
  • the formula (5) is constrained at ⁇ j ⁇ 2, wherein, ⁇ j is the index variable V j reputation provider fluctuation index, V i for the consumer, the provider V j, the minimum support count minsup ; Providers that meet the above formula (5) are classified as IOC attackers. In this way, the providers that meet the conditions of formula (5) can be classified as IOC attackers.
  • the present invention also proposes a dynamic sampling observation method to ensure flexible setting of the minsup value.
  • the support of IOC consumers and IOC providers will also increase, so it is impossible to set minsup to a static value, but minsup can be dynamically updated as the number of IOC attacks increases.
  • the method for analyzing the association relationship between IOC attackers further includes a dynamic sampling observation method, and the dynamic sampling observation method is:
  • Step 2-1 The initial minsup value is set to a preset value
  • Step 2-2 View the number of detected IOC attacks h;
  • Step 2-3 Use The function output supports the average value of the count and use it The function gets the minimum support count, uses the minimum support count to update the value of minsup, and then goes to step 2-5 to execute;
  • Step 2-4 Also use To judge whether the judgment formula is established, if not, perform steps 2-5, if the judgment formula is established, use The function gets the minimum support count, uses the minimum support count to update the value of minsup, and then goes to step 2-5 to execute;
  • Step 2-5 View the other h detected IOC attacks, use The function gets the average value of the support count, and then returns to step 2-4 for execution, where l, h, and m are natural numbers.
  • the method for deleting the unlikely IOC provider includes the following steps:
  • Step 3-2 Traverse each provider V j that belongs to ⁇ , and judge one by one whether its ⁇ j ⁇ 2, if yes, then ⁇ 1 ⁇ V j ⁇ belongs to ⁇ , where ⁇ j is the reputation of provider V j Index variables of the volatility index;
  • the goal of the present invention is to quickly detect IOC attacks and avoid TMAM overload.
  • the basic idea of the RFAA scheme for detecting IOC attacks can be described as a recursive elimination scheme to reduce suspicious providers: in the current traffic-related incident voting operations, it is unlikely to delete IOC providers. Thereby reducing the detection time of the IOC attack, the IOC attack detection process is shown in Figure 5.
  • the reputation fluctuation index can be used as the first step to delete unlikely IOC providers, ⁇ 1 represents the set of remaining providers that have not been deleted from ⁇ . in case It means that the current voting operation of traffic-related events is not attacked by the IOC, and the detection can be withdrawn.
  • V i the support count between the consumer (V i ) and each provider (V i ) in ⁇ 1 .
  • ⁇ 2 represents providers that have not been deleted from ⁇ 1 . in case Indicates that the current traffic-related voting activities are safe and free from IOC attacks. The test can be exited.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Computer Security & Cryptography (AREA)
  • Mechanical Engineering (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种车联网下安全型防御合谋攻击的系统及其方法，通过内、外合谋的方式提升自己的声誉评分来操纵TMAM。能够快速检测IOC攻击，提高车联网的安全性。通过递归消除可疑提供者和提出的声誉波动关联规则，避免了TMAM的过载。可以剥夺IOC攻击者提示其声誉评分的机会，并确保车联网中的可信信息。在不受IOC攻击者干扰的情况下，保证了TMAM的公平性和可用性。有效避免了现有技术中信誉系统可以被IOC攻击者利用来提升他们的信誉评分从而操纵TMAM的缺陷。

Description

车联网下安全型防御合谋攻击的系统及其方法
本申请要求于2019年5月20日提交中国国知局、申请号为201910418058.0的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本发明涉及车联网技术领域，本发明也涉及防御技术领域，具体涉及一种车联网下安全型防御合谋攻击的系统及其方法。
背景技术
根据中国物联网校企联盟的定义，车联网(Internet of Vehicles)是由车辆位置、速度和路线等信息构成的巨大交互网络。通过GPS、RFID、传感器、摄像头、图像处理装置等，车辆可以完成自身环境和状态信息的采集。每一个车辆都可以将这些信息传输汇聚到中央处理器，通过中央处理器的分析和处理，能够计算出不同车辆的最佳路线、及时汇报路况，从而有利于车联网系统安排信号灯周期，等等。
根据调查研究得知，预计在未来10到20年内，注册车辆的数量将达到20亿辆。由大量的基础设施和智能车辆构成的车联网将作为智能交通系统的基础。基于车联网建立的车辆对车辆(V2V)通信和车辆对基础设施(V2I)通信，能够促进车辆之间的信息合作，共享和交通相关的信息，如道路状况、交通拥堵情况等。如图1所示，通过V2V通信，一个车辆可以与附近车辆交换信息；通过V2I通信，一个车辆可以直接与路边单元(Roadside Unit，RSUs)交换信息。专用短程通信(DSRC)无线电和IEEE标准可用于车联网中的V2V和V2I通信。
车辆配备车载单元，即传感器、资源命令处理器、存储和通信设备，用于数据收集、处理和共享。在车载单元的帮助下，车辆可以自动检测与交通有关的事件，并使用车辆对车辆通信标准向他人发送警告信息。这些信息帮助车辆及时了解交通状况，从而提高交通安全效率。在与交通相关的消息交换中，每辆车都可以扮演两个角色，即提供消息的角色和接收消息的角色。
然而，车联网的高机动性、高波动性等特点使得相邻车辆之间往往互不关联，互不相识。这为恶意车辆提供了故意传播不实信息的机会。例如，恶意车辆报告一条消息，声称道路畅通。当发生交通事故或拥堵时，这些不实信息实际上极有可能危害交通安全并且降低交通系统的效率。
此外，这些不实信息不仅可能降低运输效率，还可能导致事故，甚至可能威胁到人类的生命。例如，车辆(V a)广播一条失控警告消息，警告它后面的车辆。当车辆V b接收到这条警告信息时，对于V b来说，判断这条信息的可信度并做出快速的决策是至关重要的。由于时间限制，向邻居车辆或可信的第三方求证该信息的真实性是不切实际的。如果这个警告信息是假的，那么V b刹车就会有危险。因此，在车联网中，如何有效地建立车辆之间的信任关系是非常重要的。我们希望每辆车都能检测到恶意车辆及其报告的不实信息。
信誉系统(也称为信任管理方案)使车辆能够判断接收到的消息是否可信，为网络 运营商对特定车辆的奖惩提供依据。一般情况下，车辆的信誉评分可以通过对其过去行为的评分来计算。
在信誉系统的帮助下，使用信誉系统的消费者可以收集有关某一交通事件的所有信息，并且过滤掉恶意车辆报告的不实信息，只检测出可信的信息。然后使用特定的模型将所有流量消息聚合后进行快速决策，例如:多数决定原则。交通相关消息聚合模型(TMAM)可以由中心车联网的云服务器管理，也可以由分布式车联网中的RSU管理。最后，消费者可以根据该事件的最终结果为所述消息生成评级，并上传这些评级来更新提供消息的车辆的信誉评分。
内外合谋攻击(Inside-and-outside Collusive Attack，IOC)攻击策略如图2所示，IOC攻击者对他们的信誉分数极为敏感。他们开始在约束下发起IOC攻击，而信誉评分R i表达式一般为：
R i≤ε+η 1
假设V i为攻击者之一，R i为V i的信誉评分，每个R i∈[0,1]，信誉评分阈值(ε)通常设置为一个适中的值，如0.5，当cre＝inc时，在在下述公式(1)中计算R i。R i>ε表示较高的信誉分数，这意味着车辆的消息可以被TMAM接受。这使得IOC攻击者找到一个攻击过程来提示他们的信誉评分。当R i≤ε+η 1时，V i停止伪造与交通有关的信息，并要求他身边的同谋之一发动IOC的攻击，以提升他的信誉评分。同时，其他同谋在ε+η 1≤R i≤ε+η 2的情况下可以搭便车，在相同的时间提高他们的信誉得分。此时，η 1(0≤η 1<ε)是信誉警戒线。现在要想在适当R i<ε的时候提高信誉分数已经太晚了，在这种情况下，该IOC攻击者被标记为可疑车辆，TMAM将不会相信他。这种攻击模式一直持续到R i<ε，η 2(η 1<η 2≤ε)是高信誉线。
在现有技术中，信誉系统可以分为两类:集中信誉系统和分散信誉系统。
在集中信誉系统中，所有的评级都存储在一个中央服务器中并进行处理，例如,云服务器。由于车辆通常需要在相当短的延迟内做出决策，这些集中的系统并不总是能够满足车联网严格的服务质量(QoS)要求。
在分散信誉系统中，信誉计算任务是在车辆本身或RSU中进行的。信誉评分的本地管理可能会减少与网络基础设施的交互。
在信誉系统中，最流行的一种设计是基于beta功能的。它首先计算出车辆进行的可信和恶意行为的数量。然后用beta函数计算信誉得分。
当一条信息(m i)被报告为V i时，如果m i与消费者评价的交通相关信息的真实性相同，则认为m i是可信的信息。否则，它将被认为是不实信息。对于车辆V i，信誉系统计算该车辆报告的可信消息数量(cre i)，以及该车辆报告的不实信息数量inc i。V i的信誉度得分用beta函数计算为:
Figure PCTCN2019127707-appb-000001
然而，信誉系统可以被IOC攻击者利用来提升他们的信誉评分，从而操纵TMAM。
发明内容
为解决上述问题，本发明提供了一种车联网下安全型防御合谋攻击的系统及其方法，有效避免了现有技术中信誉系统可以被IOC攻击者利用来提升他们的信誉评分从而操纵TMAM的缺陷。
为了克服现有技术中的不足，本发明提供了一种车联网下安全型防御合谋攻击的系统及其方法的解决方案，具体如下：
一种车联网下安全型防御合谋攻击的系统，包括车联网平台的云服务器或者分布式车联网中的RSU；
所述车联网下安全型防御合谋攻击的系统还包括设置在车辆上的车载单元，所述车载单元与车联网平台的云服务器或者分布式车联网中的RSU连接，所述车载单元中包括处理器所述处理器与无线通信模块连接，所述处理器还与存储器连接。
所述车联网平台的云服务器或者分布式车联网中的RSU中包括有交通相关消息聚合模型TMAM；
所述相关消息聚合模型TMAM用于在与耗流量的投票活动结束时存储来自提供者的历史消息和消费者的历史评级数据。
所述车联网下安全型防御合谋攻击的系统的方法，该方法运行在车联网平台的云服务器或者分布式车联网中的RSU上，所述方法包括如下方式：
通过信誉波动关联分析RFAA方法来检测IOC攻击，分析信誉波动特性，在当前与耗流量的事件投票操作中，从所有提供者中删除不太可能的IOC提供者。
所述信誉波动关联分析RFAA包括如下方式：
(1)支持检测IOC攻击的数据管理，所述支持检测IOC攻击的数据管理包括基于两步处理消息报告和消费者评级，所述基于两步处理消息报告和消费者评级的方法具体如下：
所述交通相关消息聚合模型TMAM应该在与耗流量的投票活动结束时存储来自提供者的历史消息和消费者的历史评级数据，所述提供者为车辆的车载单元用于提供消息，所述消费者为车辆的车载单元用于接收消息和对消息进行评级，这样就实现了基于两步处理消息报告；
每个车辆V i都被分配到一个P-C表中，该P-C表保存提供者以前在车辆上报告的消息，以及，当该车辆充当消费者时，所有提供者提供的消息和对V i的评级都存储在P-C表中，这样就实现了消费者评级；
(2)得到信誉波动关联规则，所述得到信誉波动关联规则支持检测IOC攻击，通过分析IOC攻击的特点，得到信誉波动关联规则。
所述P-C表如表1所示：
表1
Figure PCTCN2019127707-appb-000002
其中，sn表示与交通相关事件的投票行为的编号，h为车辆V i发起的与交通事件相关的投票行动数量；
Vt表示车辆V i请求与交通相关事件的投票行为时的投票时间，vt i h是车辆V i在第h次最新的投票时间，h为自然数；
P_ID(消息)表示分配给提供者的ID及其历史消息，对于V j(m j) h来说，V j是第j个提供者的ID，(m j) h是其在第h个与交通相关的事件投票操作中的消息,i、j和n为自然数且为分配给该车辆的序列号；
C_ID(评级数据)表示消费者的ID及其历史评级,对于V i(t) h来说，V i为第i个消费者的ID，(t) h为其在第h次交通相关事件投票行为中对应的真实等级，i、j和n为自然数且为分配给该车辆的序列号。
所述得到信誉波动关联规则的方法包括：
步骤1-1：把作为信誉波动指数的指标变量μ i的值初始化为0；
步骤1-2：把整型变量k的初始值设定为1；
步骤1-3：把整型变量k的值与h比较，若k>h，就结束得到信誉波动关联规则的方法；若k≤h，如果同时满足R k i≥ε+η 2、R k i≤R k-1 i、R k i连续下降且R k i≤ε+η 1，那么μ i＝μ i+1；这里，h为车辆V i发起的与交通事件相关的投票行动数量，R k i表示车辆V i在所述第k次的信誉得分，ε作为信誉评分阈值通常设置为一个0到1之间的适中的值，如0.5，η 1的取值范围为0≤η 1<ε且其作为信誉警戒线，η 2的取值范围为η 1<η 2≤ε且其作为高信誉线；
步骤1-4：如果同时满足R k i≤ε+η 1、R k i≥R k-1 i、R k i连续下降且R k i≥ε+η 2，那么μ i＝μ i+1；
步骤1-5：k＝k+1,返回步骤1-3执行。
所述信誉波动关联分析RFAA的方式还包括分析IOC攻击者之间的关联关系，所述分析IOC攻击者之间的关联关系的方法具体如下：
IOC提供者是经常出现在一起的频繁提供者，在分析IOC攻击者之间的关联关系中，索引支持数s(·)可用于标识频繁提供者；
描述出IOC消费者和IOC提供者之间的关联关系，还可以发现IOC使用者和IOC提供者经常同时出现，可以确定与IOC提供者经常同时出现的IOC消费者；
如果IOC提供者的数量超过当前与交通相关的事件投票操作中提供者的一半以上，则IOC攻击者可以成功地提升他们的信誉得分，
假设V i是一个消费者，并且Φ是当前与耗流量的事件投票操作中的一组提供者，
Figure PCTCN2019127707-appb-000003
对于每个
Figure PCTCN2019127707-appb-000004
信誉波动关联规则可以设计为如式(5)所示的：
Figure PCTCN2019127707-appb-000005
该式(5)是在μ j≥2约束下，其中，μ j为提供者V j的信誉波动指数的指标变量，V i为消费者，V j为提供者，minsup是支持计数的最小值；符合上述式(5)的提供者被区分为IOC攻击者。
所述分析IOC攻击者之间的关联关系的方法还包括动态采样观测方法，所述动态采样观测方法为：
步骤2-1：初始的minsup的值被设置为一个事先设定的值；
步骤2-2：查看检测到的IOC攻击的次数h；
步骤2-3：用
Figure PCTCN2019127707-appb-000006
函数输出支持计数的平均值并用
Figure PCTCN2019127707-appb-000007
函数得到最小支持计数，用最小支持计数来更新minsup的值，然后转到步骤2-5中执行；
步骤2-4：另外还用
Figure PCTCN2019127707-appb-000008
来判断该判断式是否成立，如果不成立，就执行步骤2-5，如果该判断式成立，就用
Figure PCTCN2019127707-appb-000009
函数得到最小支持计数，用最小支持计数来更新minsup的值，然后转到步骤2-5中执行；
步骤2-5：查看另外h次检测到的IOC攻击，用
Figure PCTCN2019127707-appb-000010
函数得到支持计数的平均值，然后返回步骤2-4中执行，其中，l、h和m为自然数。
所述删除不太可能的IOC提供者的方法包括如下步骤：
步骤3-1：初始化
Figure PCTCN2019127707-appb-000011
dr＝0,
Figure PCTCN2019127707-appb-000012
其中，Φ表示所有提供者集合，Φ1表示未从Φ中删除的提供者集合，Φ2表示未从Φ1中删除的提供者集合，dr＝0表示当前交通相关事件的投票行动是安全的，没有受到IOC攻击；
步骤3-2：遍历每个属于Φ的提供者V j，逐一判断是否其μ j≥2，如果是，那么Φ 1←{V j}属于Φ，其中，μ j为提供者V j的信誉波动指数的指标变量；
步骤3-3：如果
Figure PCTCN2019127707-appb-000013
那么遍历每个属于Φ 1的提供者V j，如果s(V i→V j)≥minsup，那么Φ 2←{V j}属于Φ 2，其中，V i表示消费者；否则dr＝0且当前交通相关事件的投票行动是安全的，没有受到IOC的攻击；
步骤3-4：如果
Figure PCTCN2019127707-appb-000014
那么
Figure PCTCN2019127707-appb-000015
否则dr＝0且当前交通相关事件的投票行动是安全的，没有受到IOC的攻击
步骤3-5：遍历每个属于Φ 1/2的V j，搜索其P-C j表，如果
Figure PCTCN2019127707-appb-000016
那么dr＝1且为检测到IOC攻击；否则，dr＝0且交通相关事件的投票行动是安全的，没有受到IOC的攻击。
本发明的有益效果为：
本发明可以修复车联网中信誉信誉系统的漏洞，IOC攻击者可以通过内、外合谋的方式提升自己的信誉评分来操纵TMAM。能够快速检测IOC攻击，提高车联网的安全性。通过递归消除可疑提供者和提出的信誉波动关联规则，避免了TMAM的过载。可以剥夺IOC攻击者提示其信誉评分的机会，并确保车联网中的可信信息。在不受IOC攻击者干扰的情况下，保证了TMAM的公平性和可用性。
附图说明
图1是车联网系统模型的示意图。
图2是IOC攻击策略的示意图。
图3是本发明的信誉波动关联分析RFAA的架构示意图。
图4是本发明的动态采样观察法的工作流程图。
图5是本发明的IOC攻击检测的流程图。
具体实施方式
下面将结合附图和实施例对本发明做进一步地说明。
如图3-图5所示，车联网下安全型防御合谋攻击的系统，包括现有的IOC攻击会对TMAM的性能造成二维损害，IOC攻击者更容易破坏TMAM的公平性和可用性。IOC消费者可以根据同谋者的信息发送评级。经过几轮攻击后，IOC攻击者可以互相帮助，快速获得较高的信誉分数，从而更容易地操作TMAM。在现有技术中，攻击者仅仅扮演提供者的角色，有时完全靠自己的诚实行为来提高自己的信誉得分。
IOC攻击可能会影响可信车辆的情绪，因为它们可能不会通过可信信息来提升自己的信誉得分。如果他们的可信信息与IOC消费者的评级不同，他们的信誉评分则不会被提升。在现有技术中，可信的信息可以使可信车辆的信誉评分得到提升。
本发明车联网下安全型防御合谋攻击的系统及其方法的核心是利用信誉波动关联分析(RFAA)的设计思想设计一种快速检测IOC攻击的防御方案。本发明与现有技术的不同之处在于:
信誉波动关联分析RFAA用于检测IOC攻击，分析信誉波动特性，在当前与耗流量的事件投票操作中，从所有提供者中删除不太可能的IOC提供者。在信誉波动关联分析RFAA方案中给出了信誉波动特征的定量方法。在本发明中，根据发现少数人不能根据多数原则改变交通相关消息聚合模型TMAM的决策。如果提供者的信誉波动特性的数量小于大多数提供者，当前与耗流量的事件投票行为是安全的，不受IOC攻击，因为不可能被少数人利用交通相关消息聚合模型TMAM。在现有技术中，聚类分析用于检测合谋攻击，不能预先发现可疑提供者是否是所有提供者中的少数。因此，现有技术将浪费更多时间来检测当前与交通相关的事件投票行为是否安全。
可以注意到，提供者很少与消费者一起出现，也很少与消费者和其他提供者一起出现。在本发明中，可以继续进行第二步来删除不太可能的IOC提供者，方法是分析消费者和每个提供者之间的关联关系(支持计数用于量化)，此时可疑提供者仍然占多数，并且具有信誉波动特性。在现有技术中，在检测合谋攻击时仅分析提供者之间的关系。
通过缩小可疑提供者的范围，可以减少数据库的搜索量，支持信誉波动关联分析RFAA方案中提出的检测IOC攻击的信誉波动关联规则。本发明在检测到IOC攻击时，应放弃当前与交通相关的事件投票行为，从而剥夺IOC攻击者提升其信誉评分的机会。在现有技术中，需要检测所有合谋攻击者，这也可能增加交通相关消息聚合模型TMAM的过载。
一种车联网下安全型防御合谋攻击的系统，包括车联网平台的云服务器或者分布式车联网中的RSU；
所述车联网下安全型防御合谋攻击的系统还包括设置在车辆上的车载单元，所述车载单元与车联网平台的云服务器或者分布式车联网中的RSU连接，所述车载单元中包括处理器，所述处理器能够是PLC、单片机、DSP处理器或者ARM处理器，所述处理器与无线通信模块连接，所述无线通信模块能够是与车联网平台连接的3G模块或者4G模块，所述处理器还与存储器连接，所述存储器能够是闪存或者外存。
所述车联网平台的云服务器或者分布式车联网中的RSU中包括有交通相关消息聚合模型TMAM；
所述相关消息聚合模型TMAM用于在与耗流量的投票活动结束时存储来自提供者的历史消息和消费者的历史评级数据。
所述车联网下安全型防御合谋攻击的系统的方法，该方法运行在车联网平台的云服务器或者分布式车联网中的RSU上，所述车联网下安全型防御合谋攻击的系统的方法包括如下方式：
通过信誉波动关联分析RFAA方法来检测IOC攻击，分析信誉波动特性，在当前与耗流量的事件投票操作中，从所有提供者中删除不太可能的IOC提供者。
其中，所述的耗流量的事件投票操作主要是指路况交互事件，即交通相关事件的投票行为。
所述信誉波动关联分析RFAA包括如下方式：
(1)支持检测IOC攻击的数据管理，所述支持检测IOC攻击的数据管理包括基于两步处理消息报告和消费者评级，所述基于两步处理消息报告和消费者评级的方法具体如下：
所述交通相关消息聚合模型TMAM应该在与耗流量的投票操作结束时存储来自提供者的历史消息和消费者的历史评级数据，而不是丢弃它们，所述提供者为车辆的车载单元用于提供消息，所述消费者为车辆的车载单元用于接收消息和对消息进行评级，这样就实现了基于两步处理消息报告；
为避免更多的搜索量耗费在检测IOC攻击上，提供者的消息应该基于区分消费者的观点，而不是作为一个整体保存在所有消费者上。每个车辆V i都可以被分配到一个P-C表中，该P-C表保存提供者以前在车辆上报告的消息，当该车辆充当消费者时，拿车辆V i举个例子，所有提供者提供的消息和对V i的评级都存储在P-C表中，这样就实现了消费者评级；
(2)得到信誉波动关联规则，所述得到信誉波动关联规则支持检测IOC攻击，通过分析IOC攻击的特点，确定信誉波动关联规则。
所述P-C表如表1所示：
表1
Figure PCTCN2019127707-appb-000017
sn表示与交通相关事件的投票行为的编号，h为车辆V i发起的与交通事件相关的投票行动数量；
vt表示车辆V i请求与交通相关事件的投票行为时的投票时间，vt i h是车辆V i在第h次最新的投票时间，h为自然数；
P_ID(消息)表示分配给提供者的ID及其历史消息，对于V j(m j) h来说，V j是第j个提供者的ID，(m j) h是其在第h个与交通相关的事件投票操作中的消息；具体来说，当V i报告“1”时，V j(m j) h被记录为V j(1) h，当V i报告“0”时，V j(m j) h被记录为V j(0) h，当V i没有报告时，V j(m j) h被记录为V j(-) h；
C_ID(评级数据)表示消费者的ID及其历史评级,对于V i(t) h来说，V i为第i个消费者的ID，(t) h为其在第h次交通相关事件投票行为中对应的真实等级，i、j和n为自然数且为分配给该车辆的序列号。
所述得到信誉波动关联规则的方法包括：
因为IOC攻击者会报告可信的消息来保持较高的信誉分数，或者报告可信的消息来操纵TMAM。IOC攻击者具有信誉波动的特征，于是就引入信誉波动指数的指标变量μ来量化IOC攻击者具有信誉波动的特征的特性，指标μ可以通过观察一辆车的信誉分数是连续增加还是下降来计算。拿V i再举个例子，R k i表示车辆V i在所述第k次的信誉得分。
步骤1-1：把作为信誉波动指数的指标变量μ i的值初始化为0，其中，μ i表示车辆V i信誉波动指数；
步骤1-2：把整型变量k的值设定为1；
步骤1-3：把整型变量k的值与h比较，若k>h，就结束得到信誉波动关联规则的方法；若k≤h，如果同时满足R k i≥ε+η 2、R k i≤R k-1 i、R k i连续下降且R k i≤ε+η 1，那么μ i＝μ i+1；R k i连续下降表示从R 1 i到R k i的值是持续下降的，这里，h为车辆V i发起的与交通事件相关的投票行动数量，R k i表示车辆V i在所述第k次的信誉得分，ε作为信誉评分阈值通常设置为一个0到1之间的适中的值，如0.5，η 1的取值范围为0≤η 1<ε且其作为信誉警戒线，η 2的取值范围为η 1<η 2≤ε且其作为高信誉线；
步骤1-4：如果同时满足R k i≤ε+η 1、R k i≥R k-1 i、R k i连续下降且R k i≥ε+η 2，那么μ i＝μ i+1；
步骤1-5：k＝k+1,返回步骤1-3执行。
通过所述得到信誉波动关联规则的方法可以在每次交通相关事件投票行动结束时执行更新μ i，为了避免在检测IOC攻击时计算μ i的冗余。对于μ i≥2来说，这意味着一名 攻击者已经发起了至少一轮的IOC攻击，以提升他的信誉评分，并报告了不实信息。当然，一些表现诚实的恶意车辆有时也能制造μ i≥2。因此，根据得到的信誉波动关联规则，可以缩小可疑提供者的筛选范围。为了有效地检测IOC攻击者，就还需要分析IOC攻击者之间的关联关系。
本发明基于以下三个特性分析IOC攻击者之间的关联关系。
共同攻击者:IOC攻击者经常一起报告消息。
角色交换:IOC攻击者可以扮演IOC消费者帮助他的同谋者扮演IOC提供者的角色。经过一轮IOC攻击后，其中一个IOC提供者将被指定为新的IOC消费者，而前IOC消费者可以成为新的IOC提供者。
多数规则:TMAM的决定应该与大多数人的意见一致。
所述信誉波动关联分析RFAA的方式还包括分析IOC攻击者之间的关联关系，所述分析IOC攻击者之间的关联关系的方法具体如下：
对于第一个特性，IOC提供者也可以是经常出现在一起的频繁提供者，在分析IOC攻击者之间的关联关系中，索引支持数s(·)可用于标识频繁提供者；例如，如果V 1,V 2,V 3是三个频繁提供者，它们的支持计数为s(V 1,V 2,V 3)，是V 1,V 2,V 3同时出现的次数，应该通过搜索P-C表来满足以下规则
s(V 1,V 2,V 3)≥minsup              (2)
其中minsup是支持计数的最小值。
对于第二个特性，描述出IOC消费者和IOC提供者之间的关联关系，还可以发现IOC消费者和IOC提供者经常同时出现；例如，如果V 6是IOC消费者和V 1,V 2,V 3是三个IOC提供者，它们之间的关联关系可以描述为V 6→{V 1,V 2,V 3}。
对于第三个特性，如果IOC提供者的数量超过当前与交通相关的事件投票操作中提供者的一半以上，IOC攻击者可以成功地提升他们的信誉得分，例如V 6也是IOC的消费者，并且V 1,V 2,V 3,V 4,V 5是当前交通相关事件投票活动的提供者。V 1,V 2,V 3可以操纵TMAM，因为它们是大多数提供者，根据关联分析规则式(3)，V 6和V 1,V 2,V 3可以被识别为IOC攻击者
s(V 6→{V 1,V 2,V 3})≥minsup         (3)
在不失一般性的前提下，假设V i是一个消费者，并且Φ是当前与耗流量的事件投票操作中的一组提供者，可以使用
Figure PCTCN2019127707-appb-000018
来表示集合IOC提供者，根据多数决定原则，对于
Figure PCTCN2019127707-appb-000019
元素的个数，应该至少是元素的一半，
Figure PCTCN2019127707-appb-000020
元素的个数可以如式(4)设置为：
Figure PCTCN2019127707-appb-000021
对于每个
Figure PCTCN2019127707-appb-000022
信誉波动关联规则可以设计为如式(5)所示的：
Figure PCTCN2019127707-appb-000023
该式(5)是在μ j≥2约束下，其中，μ j为提供者V j的信誉波动指数的指标变量，V i为消费者，V j为提供者，minsup是支持计数的最小值；符合上述式(5)的提供者被区分为IOC攻击者。这样就能把符合式(5)条件的提供者区分为IOC攻击者。
此外，本发明还提出了一种动态采样观测方法，以保证灵活设置minsup值。随着IOC多轮攻击的增加。IOC消费者和IOC提供者的支持也将增加，所以不可能将minsup设置为静态值，但是可以随着IOC攻击次数的增加动态更新minsup。
在当前交通相关事件投票行为中，如果信誉波动关联规则是可行的，则可以检测到IOC攻击。在这种情况下，当前与交通相关的事件投票、行动应该被放弃，继而剥夺了IOC攻击者提升其信誉得分的机会。
所述分析IOC攻击者之间的关联关系的方法还包括动态采样观测方法，所述动态采样观测方法为：
步骤2-1：初始的minsup的值被设置为一个事先设定的值；
步骤2-2：查看检测到的IOC攻击的次数h；
步骤2-3：用
Figure PCTCN2019127707-appb-000024
函数输出支持计数的平均值并用
Figure PCTCN2019127707-appb-000025
函数得到最小支持计数，用最小支持计数来更新minsup的值，然后转到步骤2-5中执行；
步骤2-4：另外还用
Figure PCTCN2019127707-appb-000026
来判断该判断式是否成立，如果不成立，就执行步骤2-5，如果该判断式成立，就用
Figure PCTCN2019127707-appb-000027
函数得到最小支持计数，用最小支持计数来更新minsup的值，然后转到步骤2-5中执行；
步骤2-5：查看另外h次检测到的IOC攻击，用
Figure PCTCN2019127707-appb-000028
函数得到支持计数的平均值，然后返回步骤2-4中执行，其中，l、h和m为自然数。
所述删除不太可能的IOC提供者的方法包括如下步骤：
步骤3-1：初始化
Figure PCTCN2019127707-appb-000029
dr＝0,
Figure PCTCN2019127707-appb-000030
其中，Φ表示所有提供者集合，Φ1表示未从Φ中删除的提供者集合，Φ2表示未从Φ1中删除的提供者集合，dr＝0表示当前交通相关事件的投票行动是安全的，没有受到IOC攻击；dr为判定标记，dr＝0表示当前交通相关事件的投票行动是安全的，没有受到IOC攻击；dr＝1表示存在IOC攻击。
步骤3-2：遍历每个属于Φ的提供者V j，逐一判断是否其μ j≥2，如果是，那么Φ 1←{V j}属于Φ，其中，μ j为提供者V j的信誉波动指数的指标变量；
步骤3-3：如果
Figure PCTCN2019127707-appb-000031
那么遍历每个属于Φ 1的提供者V j，如果s(V i→V j)≥minsup，那么Φ 2←{V j}属于Φ 2，其中，V i表示消费者；否则dr＝0且当前交通相关事件的投票行动是安全的，没有受到IOC的攻击；
步骤3-4：如果
Figure PCTCN2019127707-appb-000032
那么
Figure PCTCN2019127707-appb-000033
否则dr＝0且当前交通相关事件的投票行动是安全的，没有受到IOC的攻击；
步骤3-5：遍历每个属于Φ 1/2的V j，搜索其P-C j表，如果
Figure PCTCN2019127707-appb-000034
那么dr＝1且为检测到IOC攻击；否则，dr＝0且交通相关事件的投票行动是安全的，没有受到IOC的攻击。
在信誉波动关联规则的支持下，本发明的目标是快速检测IOC攻击，避免TMAM过载。实际上检测IOC攻击的RFAA方案的基本思想可以描述为递归消除方案，以减少可疑提供者:在当前与交通相关的事件投票操作中，不太可能删除IOC提供者。从而减少了IOC攻击的检测时间，IOC攻击检测过程如图5所示。
可以使用信誉波动指数作为删除不太可能的IOC提供者的第一步，Φ 1表示未从Φ中删除的其余提供者的集合。如果
Figure PCTCN2019127707-appb-000035
则表示当前与交通相关的事件投票操作不受IOC攻击，检测可以退出。
通常，如果一个提供者很少与V i一起出现，那么他也很少与V i和其他提供者一起出现。因此，我们可以继续第二步删除不太可能的IOC提供者，方法是在Φ 1中分析消费者(V i)和每个提供者(V i)之间的支持计数。这里Φ 2表示未从Φ 1中删除的提供者。如果
Figure PCTCN2019127707-appb-000036
表示当前与交通相关的投票活动是安全的，不受IOC攻击。检测可以退出。
经过这两个步骤，可以明显减少P-C表的搜索量。对于每一个
Figure PCTCN2019127707-appb-000037
我们可以利用信誉波动关联规则搜索IOC的P-C表来检测IOC攻击。Φ、Φ 1和Φ 2为集合变量。
以上以用实施例说明的方式对本发明作了描述，本领域的技术人员应当理解，本公开不限于以上描述的实施例，在不偏离本发明的范围的情况下，可以做出各种变化、改变和替换。

Claims (9)

  1. 一种车联网下安全型防御合谋攻击的系统，其特征在于，包括车联网平台的云服务器或者分布式车联网中的RSU；
    所述车联网下安全型防御合谋攻击的系统还包括设置在车辆上的车载单元，所述车载单元与车联网平台的云服务器或者分布式车联网中的RSU连接，所述车载单元中包括处理器，所述处理器与无线通信模块连接，所述处理器还与存储器连接。
  2. 根据权利要求1所述的车联网下安全型防御合谋攻击的系统，其特征在于，所述车联网平台的云服务器或者分布式车联网中的RSU中包括有交通相关消息聚合模型TMAM；
    所述相关消息聚合模型TMAM用于在与耗流量的投票活动结束时存储来自提供者的历史消息和消费者的历史评级数据。
  3. 一种车联网下安全型防御合谋攻击的方法，其特征在于，该方法运行在车联网平台的云服务器或者分布式车联网中的RSU上，所述方法包括如下方式：
    通过信誉波动关联分析RFAA方法来检测IOC攻击，分析信誉波动特性，在当前与耗流量的事件投票操作中，从所有提供者中删除不太可能的IOC提供者。
  4. 根据权利要求3所述的车联网下安全型防御合谋攻击的方法，其特征在于，所述信誉波动关联分析RFAA包括如下方式：
    (1)支持检测IOC攻击的数据管理，所述支持检测IOC攻击的数据管理包括基于两步处理消息报告和消费者评级，所述基于两步处理消息报告和消费者评级的方法具体如下：
    交通相关消息聚合模型TMAM在与耗流量的事件投票操作结束时，存储来自提供者的历史消息和消费者的历史评级数据，所述提供者为车辆的车载单元用于提供消息，所述消费者为车辆的车载单元用于接收消息和对消息进行评级，以实现基于两步处理消息报告；
    每个车辆V i都被分配到一个P-C表中，该P-C表保存提供者以前在车辆上报告的消息，以及，当该车辆充当消费者时，所有提供者提供的消息和对V i的评级以实现消费者评级；
    (2)得到信誉波动关联规则，所述得到信誉波动关联规则支持检测IOC攻击，通过分析IOC攻击的特点，得到信誉波动关联规则。
  5. 根据权利要求4所述的车联网下安全型防御合谋攻击的方法，其特征在于，所述P-C表如表1所示：
    表1
    Figure PCTCN2019127707-appb-100001
    其中，sn表示与交通相关事件的投票行为的编号，h为车辆V i发起的与交通事件相关的投票行动数量；
    vt表示车辆V i请求与交通相关事件的投票行为时的投票时间，vt i h是车辆V i在第h次最新的投票时间，h为自然数；
    P_ID(消息)表示分配给提供者的ID及其历史消息，对于V j(m j) h来说，V j是第j个提供者的ID，(m j) h是其在第h个与交通相关的事件的投票操作中的消息,i、j和n为自然数且为分配给该车辆的序列号；
    C_ID(评级数据)表示消费者的ID及其历史评级,对于V i(t) h来说，V i为第i个消费者的ID，(t) h为其在第h次交通相关事件投票行为中对应的真实等级，i、j和n为自然数且为分配给该车辆的序列号。
  6. 根据权利要求4所述的车联网下安全型防御合谋攻击的方法，其特征在于，所述得到信誉波动关联规则的方法包括：
    步骤1-1：把作为信誉波动指数的指标变量μ i的值初始化为0；
    步骤1-2：把整型变量k的初始值设定为1；
    步骤1-3：把整型变量k的值与h比较，若k>h，就结束得到信誉波动关联规则的方法；若k≤h，如果同时满足R k i≥ε+η 2、R k i≤R k-1 i、R k i连续下降且R k i≤ε+η 1，那么μ i＝μ i+1；这里，h表示车辆V i发起的与交通事件相关的投票行动数量，R k i表示车辆V i在所述第k次的信誉得分，ε作为信誉评分阈值通常设置为一个0到1之间的适中的值，η 1的取值范围为0≤η 1<ε且其作为信誉警戒线，η 2的取值范围为η 1<η 2≤ε且其作为高信誉线；
    步骤1-4：如果同时满足R k i≤ε+η 1、R k i≥R k-1 i、R k i连续下降且R k i≥ε+η 2，那么μ i＝μ i+1；
    步骤1-5：k＝k+1,返回步骤1-3执行。
  7. 根据权利要求6所述的车联网下安全型防御合谋攻击的方法，其特征在于，所述通过信誉波动关联分析RFAA方法来检测IOC攻击的方式还包括分析IOC攻击者之间的关联关系，所述分析IOC攻击者之间的关联关系的方法具体如下：
    IOC提供者是经常出现在一起的频繁提供者，在分析IOC攻击者之间的关联关系中，索引支持数s(·)可用于标识频繁提供者；
    描述出IOC消费者和IOC提供者之间的关联关系，可以确定与IOC提供者经常同时出现的IOC消费者；
    如果IOC提供者的数量超过当前与交通相关的事件投票操作中提供者的一半以上，则IOC攻击者可以成功地提升他们的信誉得分，
    如果V i是一个消费者，并且Φ是当前与耗流量的事件投票操作中的一组提供者，
    Figure PCTCN2019127707-appb-100002
    Figure PCTCN2019127707-appb-100003
    对于每个
    Figure PCTCN2019127707-appb-100004
    信誉波动关联规则可以设计为如式(5)所示的：
    Figure PCTCN2019127707-appb-100005
    该式(5)是在μ j≥2约束下，其中，μ j为提供者V j的信誉波动指数的指标变量，V i为消费者，V j为提供者，minsup是支持计数的最小值；符合上述式(5)的提供者被区分为IOC攻击者。
  8. 根据权利要求7所述的车联网下安全型防御合谋攻击的方法，其特征在于，所述分析IOC攻击者之间的关联关系的方法还包括动态采样观测方法，所述动态采样观测方法为：
    步骤2-1：初始的minsup的值被设置为一个事先设定的值；
    步骤2-2：查看检测到的IOC攻击的次数h；
    步骤2-3：用
    Figure PCTCN2019127707-appb-100006
    函数输出支持计数的平均值并用
    Figure PCTCN2019127707-appb-100007
    函数得到最小支持计数，用最小支持计数来更新minsup的值，然后转到步骤2-5中执行；
    步骤2-4：另外还用
    Figure PCTCN2019127707-appb-100008
    来判断该判断式是否成立，如果不成立，就执行步骤2-5，如果该判断式成立，就用
    Figure PCTCN2019127707-appb-100009
    函数得到最小支持计数，用最小支持计数来更新minsup的值，然后转到步骤2-5中执行；
    步骤2-5：查看另外h次检测到的IOC攻击，用
    Figure PCTCN2019127707-appb-100010
    函数得到支持计数的平均值，然后返回步骤2-4中执行，其中，l、h和m为自然数。
  9. 根据权利要求7或8所述的车联网下安全型防御合谋攻击的方法，其特征在于，所述删除不太可能的IOC提供者的方法包括如下步骤：
    步骤3-1：初始化
    Figure PCTCN2019127707-appb-100011
    其中，Φ表示所有提供者集合，Φ1表示未从Φ中删除的提供者集合，Φ2表示未从Φ1中删除的提供者集合，dr＝0表示当前交通相关事件的投票行动是安全的，没有受到IOC攻击；
    步骤3-2：遍历每个属于Φ的提供者V j，逐一判断是否其μ j≥2，如果是，那么Φ 1←{V j}属于Φ，其中，μ j为提供者V j的信誉波动指数的指标变量；
    步骤3-3：如果
    Figure PCTCN2019127707-appb-100012
    那么遍历每个属于Φ 1的提供者V j，如果s(V i→V j)≥minsup，那么Φ 2←{V j}属于Φ 2，其中，V i表示消费者；否则dr＝0且当前交通相关事件的投票行动是安全的，没有受到IOC的攻击；
    步骤3-4：如果
    Figure PCTCN2019127707-appb-100013
    那么
    Figure PCTCN2019127707-appb-100014
    否则dr＝0且当前交通相关事件的投票行动 是安全的，没有受到IOC的攻击；
    步骤3-5：遍历每个属于Φ 1/2的V j，搜索其P-C j表，如果
    Figure PCTCN2019127707-appb-100015
    那么dr＝1且为检测到IOC攻击；否则，dr＝0且交通相关事件的投票行动是安全的，没有受到IOC的攻击。
PCT/CN2019/127707 2019-05-20 2019-12-24 车联网下安全型防御合谋攻击的系统及其方法 WO2020233114A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
US17/612,720 US12010518B2 (en) 2019-05-20 2019-12-24 System and method for securely defending against collusive attack under internet of vehicles
JP2022516253A JP7407913B2 (ja) 2019-05-20 2019-12-24 車両のインターネットにおける共謀攻撃に対する安全な防御のためのシステム及びその方法

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910418058.0 2019-05-20
CN201910418058.0A CN110139278B (zh) 2019-05-20 2019-05-20 车联网下安全型防御合谋攻击的系统的方法

Publications (1)

Publication Number Publication Date
WO2020233114A1 true WO2020233114A1 (zh) 2020-11-26

Family

ID=67571296

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/127707 WO2020233114A1 (zh) 2019-05-20 2019-12-24 车联网下安全型防御合谋攻击的系统及其方法

Country Status (3)

Country Link
JP (1) JP7407913B2 (zh)
CN (1) CN110139278B (zh)
WO (1) WO2020233114A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110139278B (zh) * 2019-05-20 2020-08-04 西安安盟智能科技股份有限公司 车联网下安全型防御合谋攻击的系统的方法
CN111696352A (zh) * 2020-06-09 2020-09-22 北京百度网讯科技有限公司 交通信息的处理方法、装置、电子设备及存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103167019A (zh) * 2013-02-01 2013-06-19 中国科学技术大学 一种车载自组网中的混合式信任系统及方法
CN104717229A (zh) * 2015-04-01 2015-06-17 江苏大学 车联网中可抵御多种虚假身份来源的Sybil攻击的方法
CN107431695A (zh) * 2015-03-06 2017-12-01 诺基亚技术有限公司 用于在线投票系统中的互助共谋攻击检测的方法和装置
US20180115898A1 (en) * 2016-10-25 2018-04-26 Samsung Electronics Co., Ltd. Method of authenticating external vehicle and vehicle capable of performing same
CN108400973A (zh) * 2018-02-02 2018-08-14 中原工学院 车载自组织网络中基于交通流模型的虚假消息检测方法
CN110139278A (zh) * 2019-05-20 2019-08-16 西安安盟智能科技股份有限公司 车联网下安全型防御合谋攻击的系统及其方法

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101345627B (zh) * 2008-08-12 2011-02-16 中国科学院软件研究所 一种p2p网络中基于行为相似度的共谋团体识别方法
US8606831B2 (en) 2011-07-08 2013-12-10 Georgia Tech Research Corporation Systems and methods for providing reputation management
CN103294558B (zh) * 2013-05-29 2016-09-21 北京大学 一种支持动态信任评估的MapReduce调度方法
CN104836801A (zh) * 2015-04-22 2015-08-12 上海物联网有限公司 一种基于Binomial的无线传感器网络信任管理方法
CN105809032B (zh) * 2016-03-11 2018-10-16 上海师范大学 社交网络中面向信誉系统的共谋攻击模糊探测方法
CN106953839B (zh) * 2017-01-13 2020-06-16 重庆邮电大学 车联网中非可信资源传播的阻控系统及方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103167019A (zh) * 2013-02-01 2013-06-19 中国科学技术大学 一种车载自组网中的混合式信任系统及方法
CN107431695A (zh) * 2015-03-06 2017-12-01 诺基亚技术有限公司 用于在线投票系统中的互助共谋攻击检测的方法和装置
CN104717229A (zh) * 2015-04-01 2015-06-17 江苏大学 车联网中可抵御多种虚假身份来源的Sybil攻击的方法
US20180115898A1 (en) * 2016-10-25 2018-04-26 Samsung Electronics Co., Ltd. Method of authenticating external vehicle and vehicle capable of performing same
CN108400973A (zh) * 2018-02-02 2018-08-14 中原工学院 车载自组织网络中基于交通流模型的虚假消息检测方法
CN110139278A (zh) * 2019-05-20 2019-08-16 西安安盟智能科技股份有限公司 车联网下安全型防御合谋攻击的系统及其方法

Also Published As

Publication number Publication date
JP2022533803A (ja) 2022-07-25
US20220279352A1 (en) 2022-09-01
CN110139278B (zh) 2020-08-04
CN110139278A (zh) 2019-08-16
JP7407913B2 (ja) 2024-01-04

Similar Documents

Publication Publication Date Title
WO2020103524A1 (zh) 一种车联网的异常行为检测方法、装置和系统
CN111680745B (zh) 一种基于多源交通大数据融合的突发拥堵判别方法及系统
Mashhadi et al. Quality control for real-time ubiquitous crowdsourcing
CN107567005B (zh) 基于人工免疫系统的车联网车辆异常行为检测方法及系统
CN106887137B (zh) 拥堵事件提示方法及装置
CN108428341A (zh) 一种基于人机融合的交通应急管理调度方法及系统
CN106373396A (zh) 基于智能交通云控制系统的控制服务器
WO2020233114A1 (zh) 车联网下安全型防御合谋攻击的系统及其方法
CN114827198B (zh) 一种应用于车联网的多层中心异步联邦学习方法
Kong et al. Multimedia data fusion method based on wireless sensor network in intelligent transportation system
CN106971535B (zh) 一种基于浮动车gps实时数据的城市交通拥堵指数计算平台
CN109361577A (zh) 一种基于分布式传感器的数据监测方法和系统
CN114301935B (zh) 一种基于声誉的物联网边云协同联邦学习节点选择方法
CN114418109A (zh) 微服务架构下联邦学习的节点选择与聚合优化系统及方法
Carli et al. Monitoring traffic congestion in urban areas through probe vehicles: A case study analysis
Liu et al. Internet of things based solutions for transport network vulnerability assessment in intelligent transportation systems
Zhao et al. Fedmix: A sybil attack detection system considering cross-layer information fusion and privacy protection
US12010518B2 (en) System and method for securely defending against collusive attack under internet of vehicles
CN104008176A (zh) 基于分布式处理和消息驱动模式的业务处理方法
Wu et al. A trusted paradigm of data management for blockchain-enabled Internet of Vehicles in smart cities
JP7440065B2 (ja) ブロックチェーンネットワークシステム
Codeca et al. Traffic routing in urban environments: The impact of partial information
Wang et al. A Reputation Management Framework for MANETs
Roccetti et al. Modeling and experimenting with vehicular congestion for distributed advanced traveler information systems
Abbasi et al. A Blockchain-enabled edge supported e-challan mechanism for content-centric Internet of Vehicles

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19929906

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022516253

Country of ref document: JP

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19929906

Country of ref document: EP

Kind code of ref document: A1