Design of a Blockchain-Based Traceability System with a Privacy-Preserving Scheme of Zero-Knowledge Proof

With the fast development of the industrial Internet, its interconnectivity poses new challenges for the cooperation of industrial entities. Cooperation among these entities is built on trust, and trust is based on high-quality industrial products at reasonable prices. A traceability system can play an essential role in objectively reﬂecting the production process and promoting this trust. However, traditional traceability systems often have data privacy issues. Because traceability data are collected or generated during the production process (namely, production-related data), they could be considered privacy data. Several researchers have introduced privacy protection schemes into the traceability system, such as authentication or encryption. Nevertheless, when a privacy protection scheme is established, the original data are disclosed to the legal user of the system, but the data may still be leaked intentionally or unintentionally. Except for data privacy issues, a traditional traceability system can be vulnerable to network attacks, data unavailability, and reliability issues. The authors conducted a study to overcome these shortcomings, and this paper reports the results. We built a traceability prototype system using a blockchain protocol and a zero-knowledge proof method. First, we built a blockchain to record key production process data, aiming to maintain data reliability and availability. Second, through an analysis of traceability purpose using production knowledge, the traceability purpose could be divided into multiple provable statements. By introducing privacy protection through a zero-knowledge proof, the traceability process was converted to proving relative statements. Finally, the statements were validated by a smart contract that provided openness and reliability during the traceability process. Analysis has shown that our approach could meet the requirements for high security and privacy. In addition, the paper also discusses the calculation cost of the traceability process to show our work’s viability. The traceability system described in this paper creates new possibilities for constructing a healthy and reliable trust relationship between production entities to provide further support in the development of the industrial Internet.


Introduction
With the application of information technology (IT) to traditional industrial production, production power has significantly increased, and senior automation and information technology have optimized the production process. However, as the scale of production expands, the industrial production model may be overwhelmed by high production levels, the bullwhip effect, or biased pricing [1]. ese issues impair the trust and cooperation between industry entities [2]. To break the production limit, Industry 4.0 [3] and the industrial Internet [4] were introduced by Germany in 2010 and General Electric in 2012. e primary purpose of this structure is to connect people, data, and machines with an open and globalized network and to achieve a high degree of integration of industrial systems with computing, analysis, and IT systems [5]. In particular, the industrial Internet may create a data corridor between each element in production scenarios. By integrating the traditional manufacturing technology with big data analysis, artificial intelligence, and advanced semiconductor technology, industrial Internet could reform the entire production and cooperation model [6].
In traditional industrial production, cooperation among entities depends on the supply chain [7], and the traceability data are one of the key parts of the supply chain system to ensure the high efficiency and stable operation. Traceability data are also an important constituent for improving the quality of industrial products [8] and optimizing the supply chain and the production process [9]. As an important part of the supply chain, a traceability system can significantly affect product quality control, order management, and production service. However, with the continuous expansion of the supply chain and development of industrial Internet, the traceability system may face challenges due to the requirement of interconnection and open cooperation between industrial entities [10]. Current traceability systemrelated data are not sensitive. However, some field data related to the production process can be unsafe for a company to publish due to industrial confidentiality. Against an industrial background, sensitive production data can objectively reflect the quality and even advanced technology of industrial products, so traceability requirements for these types of sensitive data do exist. Furthermore, tracing these production data can promote and encourage interindustry cooperation and interactions.
is new type of cooperation and production may transform the dominant industrial model.
For regular scenarios, traceability can be treated as a process that satisfies a particular purpose related to demand. In this situation, the initiator of traceability should have a clear purpose, such as locating quality problems in the production process [11] or verifying whether a particular process in the production process satisfies required standards or specifications. No matter the purpose of the traceability, it is ultimately up to the traceability initializer to judge the relevant data obtained from a traceability system. However, acquirers of traceability data might lack knowledge about production in an actual traceability process. After the related data are received, a third party is needed to interpret the result using traceability data, which may eventually cause production data to leak. erefore, the traceability process for sensitive productionrelated data should avoid data transmission.
To achieve the above, this study adopts a privacy protection mechanism based on a zero-knowledge proof and realizes the traceability of the target with no need for the traceability data owner to provide any original data. In addition, the traceability system still needs to solve the problem of original data availability and reliability, and the proof should be open and without the possibility of repudiation. To satisfy the above requirements, this paper introduces blockchain technology into the design of traceability system, which could provide features that are tamper-resistant, unable to be repudiated, and open to supervision [12]. To design a complete system, this study made the following assumptions: (1) Raw industrial production data are stored in the respective production domains, while the related data digest is stored in the blockchain. (2) e traceability process was initialized by the traceability data acquirer with a clear purpose or an expected traceability result. (3) ere is a correlation between the traceability purpose and the industrial traceable production data.
Under the above premise, this paper introduces a zeroknowledge proof for raw data production.
rough a purpose analysis of traceability data acquirer, the production process was converted into multiple statements and adopted a zero-knowledge proof engine generating a validator to prove those statements. Finally, through publishing a smart contract, the traceability process is completed in an open and fair manner. e innovations of this paper are as follows: (1) An abstraction of the industrial production process into several traceability features according to their traceability is developed. (2) A privacy-preserving traceability system architecture for production data is constructed and the algorithm flow involved in the architecture is explained. (3) Availability and security issues are discussed through a comparative analysis.

Blockchain and Distributed Ledgers.
A blockchain is a type of distributed ledger system designed on a cryptography algorithm, peer-to-peer (P2P) network, and distributed consensus algorithm [13]. Blockchain has attracted much attention since its creation. Many researchers have shown increasing interest in the application of blockchain. Generally speaking, blockchain can be divided into two categories: unauthorized blockchain and authorized blockchain [14]. e former category is usually used to build payment systems instead of centralized banks, such as Bitcoin or Ethereum [15,16], and the latter is designed for specific application scenarios such as medical, agrifood, or other fields. No matter what type of blockchain is being used, it can help to build trust between participants in an exchange. Authorized blockchain supported with smart contracts could be applied to financial, medical, and logistics scenarios [17]. Moreover, its features of tamper-resistance and nonrepudiation may provide highly reliable data that could be the basis for industrial entities creating cooperative relationships. Especially in the research area of combining blockchain with industrial Internet or industry 4.0, the security and trust features of blockchain [18] may help industrial entities create healthy cooperative relationships and promote the production level. e main idea of the traceability system designed in this paper is to trace the privacy data generated in the production process. In industrial production scenarios, such as the industrial Internet of things or the industrial Internet, the data generated from billions of sensors, controllers, and data collectors makes it possible for entities to make production more intelligent, optimize production plans, and realize cooperative production [19]. To create a reliable traceability system, the first step is to guarantee the reliability and availability of production data [20]. In traditional centralized traceability approaches, there are many potential information security issues, including denial-of-service attacks, spoofing attacks, and data leaking and tampering, while a blockchain-based approach may be immune to the security issues above and make the data both tamper-resistant and highly available [21]. In this situation, the circulation of data is treated as a transaction, and data digests can be recorded in a transaction for confirmation by all participants. Supported by blockchain, a traceability system able to fully record industrial production could be created.
Nevertheless, the openness of blockchain may also create privacy issues [22]. Production-related data may be tightly bonded with industrial secrets and sensitive data, making it impossible for entities to share their production data for traceability. erefore, a privacy-preserving scheme should be deployed.

Zero-Knowledge Proofs.
Zero-knowledge proofs can enable one subject to verify the correctness of a statement put forward by another subject without involving any raw data or relying on a third party [23]. erefore, zeroknowledge proofs can be used as effective privacy protection mechanisms.
ere are two leading roles involved in the zero-knowledge proof process. e first is the prover, who declares a statement and generates a proof with raw data. e second role is the verifier, or the proof receiver, who has the ability to verify the proof. Zero-knowledge proofs are widely used in privacy-preserving schemes due to their completeness, soundness, and zero-knowledge [24]. Completeness means the statement can be verified by the prover and convince the verifier of its veracity. Its soundness provides an environment in which the prover cannot cheat the verifier with a false proof. Zero-knowledge ensures that the raw data is never revealed to the public. e prover can always maintain their ownership of the raw data during the proving and verifying processes to protect their privacy.
In the real-world usage of zero-knowledge proof schemes, a toolkit based on zero-knowledge succinct noninteractive arguments of knowledge (zkSNARK) is introduced to build corresponding systems [25]. zkSNARK allows the prover to prove its statement with low process complexity using a simple message. erefore, the toolkit based on zkSNARK is widely used in the design of blockchainbased applications [26]. According to the application scenario, a zkSNARK-based toolkit offers a flexible and effective way to create a smart contract for automatic verification and generate the proof with raw data.
In this situation, the application scenarios of zeroknowledge proof are significantly expanded and provide the possibility for the implementation of traceability in this paper.

Related Work.
Research on the combination of blockchain and traceability systems has shown that the data recorded in a blockchain can provide reliable support for data traceability so long as production data can be stored with high reliability. Previous work has successfully connected the traceability process with the data produced during production [27]. e traceability of production data is beneficial for tracking production drawbacks and raising production quality [28], and, in the research area of traceability, Chen et al. [29] demonstrated the relationship between quality control and traceability and then designed a quality control model based on traceability. On this basis, Tsai and Wang et al. [30] then designed a cooperative production method based on the production data to improve and optimize the production process. By importing blockchain into traceability system design, more researchers have concentrated on the design of blockchain-based decentralized traceability systems. Helo et al. [31] designed a high-performance traceability model for the supply chains based on blockchain, Radio Frequency Identification (RFID), and Internet of ings (IoT) technology. Zhu et al. [32] optimized the supply chain by using blockchain-based traceability system to trace production processes and coordination. Xiao et al. [33] designed a traceability model for the agrifood industry, providing a safe and traceable environment for food production quality control and anticounterfeiting. Tarun [34], based on blockchain, constructed a traceability system for textile manufacturing and improved production efficiency. Uddin [35] built a blockchain-based traceability framework for the pharmacy industry that provided reliability verification for the circulation of medicine. Patelli et al. [36] built a traceability system for the supply chain management of food industry based on blockchain, which is immune to several network attacks compared to the traditional traceability mechanism. e above studies have proven that blockchain technology can be effectively integrated with traceability mechanisms to improve production efficiency and product quality to ensure data reliability. Except for studies on the reliability of traceability data, privacy-preserving schemes for traceability data were also discussed by researchers. Yang [37] used RFID encryption to provide production data privacy during data collection. Wang [38] treated the traceability process as transactions in the blockchain and divided the traceability process into three actions: demanding, pricing, and trading.
Privacy-preserving mechanisms have also been introduced to avoid privacy leakage issues formed by the openness of blockchain. e above research shows that privacy preservation methods mainly depend on access control, identity authorization, or data encryption. Although the traceability process is protected, raw traceability data can still be leaked by the traceability data receiver.
In this paper, a zero-knowledge proof is used to protect the raw traceability data. Zero-knowledge proofs can provide proof for satisfying specified conditions in a specific scene without disclosing any private information. Currently, zero-knowledge proofs are widely used in digital currency. Zcash realized a privacy-protected digital currency system by applying a zero-knowledge proof. In addition, Eberhardt [39] combined a zero-knowledge proof with an Ethereum smart contract by constructing ZoKrates, realizing a zeroknowledge proof mode of offline computing and online verification, thus expanding the possible applications of zero-knowledge proofs. Based on this, Westerkamp [40] designed a side chain proof mechanism using ZoKrates. In addition to the field of digital currency, Ibrahem [41] applied zero-knowledge proofs and homomorphic encryption to an anonymous voting system. Rasheed et al. [42] realized an anonymity authentication method based on the premise of protecting user privacy by applying a zero-knowledge proof in the area of Internet of vehicle. Jeong et al. [43] applied smart contracts and zero-knowledge proofs in online real Security and Communication Networks estate transactions to provide a transaction process with a privacy protection mechanism. Qi et al. [44] applied a zeroknowledge proof to the auto insurance industry, achieving an effective insurance evaluation method based on usage habits while preserving privacy. Umar et al. [45] combined zero-knowledge proofs with wireless body sensors to effectively avoid privacy leakage caused by malicious attackers monitoring communication channels. Huang et al. [46] proposed an auditable information-sharing mechanism for the industrial Internet based on a zero-knowledge proof mechanism, which effectively avoided the leakage of sensitive information into the industrial environment. e above research results show that zero-knowledge proofs can provide a proof process with a privacy protection mechanism for different fields according to their needs, and so it is feasible and beneficial to attempt to apply it to the privacy protection process of production data traceability.

Architecture Review.
Based on industrial production data, as industrial production entities record production data digests in the blockchain, these production data can objectively describe the production process and reflect the technologies of production and the flow of the production process. Traceability processes are often used to show the high quality or advanced technology of production [47]. In this situation, the traceability process needs to be open and fair. However, the traceability data acquirers may lack production-related knowledge and cannot judge whether acquired data could reach their purpose. erefore, the current solution relies on a trusted third party, which means that the traceability data acquirers need to inform their purposes to a trusted third party, entrust it as an agent to receive the traceability data, and make a final judgement [48]. However, this solution still exposes the production data to a third-party verifier. At the same time, the verification process would not be open and transparent. To establish an open, fair, and privacy-preserving traceability mechanism, this paper designed a new traceability system based on a zero-knowledge proof combined with a smart contract and blockchain, which is called a zero-knowledge-based traceability system (ZKTS), to create a privacy data traceability method independent from any third party in order to protect the privacy of producers. e architecture diagram of the traceability system is shown in Figure 1.
e traceability system in this paper contains three layers. e first is the physical data layer, which contains the data produced by each independent entity according to the actual production, the physical data generated by the transportation and sales process, and the raw data saved in the database and constructed by an independent production entity. At the same time, the digest of raw data was published in an authorized blockchain system called a data chain. Data chain is open for all certificated users to access the data digest of raw data. e second is the data privacy layer, composed of a zero-knowledge engine and its related external interfaces. e data privacy layer is mainly responsible for receiving the traceability features and the related data digest from the upper layer, building the smart contract, and interacting with the data provider to generate the proof for traceability using a zero-knowledge proof engine. e data privacy layer is responsible for the core function of the traceability system, privacy preservation. e third layer is the application layer; in this paper, we define the purpose of the traceability data acquirer as the traceability purpose, and the application layer maintains the authentication process and the traceability purpose analysis process or the traceability feature-generating algorithm. is algorithm is used to analyze the primary purpose of traceability and generate related features and could satisfy the traceability purpose. As shown in Figure 1, the entire system contains two relatively independent blockchains. One is used to ensure the integrity and reliability of relevant production data, and the other is used for traceability verification in public scenarios. e two blockchains cannot interact directly but can be accessed through authentication with supervision. e specific functions of each layer in the traceability model are as follows.

Physical Data Layer.
e physical data layer includes sensors, controllers, data collection devices, and other production-related devices. In the physical data layer, production data can be mapped to devices using a unique label through RFID or other technologies. e physical data layer collects the data generated by those production devices, and these collected data can be used to generate traceability features. In this study, we assumed that no fake data had been created in the physical data layer and that the data digest would be published in a particular blockchain system (data chain).

Data Privacy Layer.
e data privacy layer is the key component ensuring the protection of privacy in the traceability system. e data privacy layer has a data extraction and privacy processing engine. First, the data privacy layer collects the data digest of production from the data chain. According to the analysis of the purpose of the traceability data acquirer, the privacy processing engine then generates a smart contract with need-to-proof issues. A witness is then generated to interact with the owner of the raw production data to generate the proof. e processing of the data privacy layer involves the privacy information of production, so the process is offline. After the proof is generated, the smart contract, related proof, and traceability features are disclosed to verify whether traceability is achieved.

Traceability Application Layer.
e traceability application layer directly interacts with the traceability data acquirer and production data owner. e primary function of the application layer is to provide a traceability interface for both sides with an authentication mechanism and also to provide a platform for mutual traceability negotiation. e traceability application layer was established by industrial entities and traceability-related individuals or entities. e traceability application was developed with a smart contract on a public or light-authenticated blockchain system. e traceability application layer can publish the smart contract generated by the data privacy layer and provide a veri cation interface for the proof. A blockchain-based application layer is an essential part of the traceability system, allowing the system to be both fair and transparent.

Participant
Design. In the traceability system we designed, given the ownership of the traceability data, there were three roles: the initializer of the traceability process, called the traceability data acquirer (TDA); the traceability data owner, called the traceability data provider (TDP); and a third party, called the traceability agent (TA). e TDP was the original holder and provider of the production data to be traced. In production-related traceability, the data has production technology privacy that needs to be protected during the production and circulation of the relevant products. e TDA was the initializer of the tracing process and could be a partner or a consumer with either a cooperative or a transaction-based relationship with the TDP. e TDA initiated the tracing process with a clear purpose such as judgement of production quality, making it a sign of traceability completion. e TA introduced in this paper is a new role. e TA held the production knowledge of the TDP, which means that the TA could analyze the primary purposes of the TDA. According to the features generated by the TA, smart contracts with related need-toproof issues were created and transferred to TDP. According to those need-to-proof issues, the TDP generates the proof with raw production data and publishes it to the public.
Since the behavior of the TA did not involve the privacy of both sides of the traceability process, the whole traceability process can be conducted transparently to ensure the openness and e ectiveness.

Traceability Process Design.
e traceability process designed in this study contains four phases: the authentication phase, preprocessing phase, construction phase, and veri cation phase. In di erent stages, di erent roles may execute corresponding work ows, and a complete process of our approach together with the relationship between roles and work ows is shown in Figure 2.
A typical traceability process is shown in Figure 2. In the authentication phase, an authentication scheme is implemented based on the data recorded in the data chain. A traceability request is then sent from the TDA to the TA. In the preprocessing phase, the TA extracts the tracing proposal from the request, generates the traceability features, and transfers them to the TDP. In the construction phase, the TDP receives the traceability features and generates the proof using the raw production data. Meanwhile, a smart contract is published by the TA for veri cation. Finally, in the veri cation phase, the TA receives the generated proof and passes it to the TDA. After the complete process, the TDA determines whether traceability has been achieved. A more detailed explanation of the algorithm and its process is fully explained in what follows.

Authentication Phase.
In this phase, due to the openness of the traceability system, the TDA and TDP both

Security and Communication Networks
have the privilege of accessing the data digest stored in the data chain, which means the TDA and TDP each have a public and private key pair: PKa/SKa and PKp/SKp, respectively. Furthermore, in the process of production trading, a relationship between the TDA and TDP is established and the related transaction data are also recorded in the data chain; these data could be used to generate ProductData (PD) using the following formula: ProductData PID|TxTS|TxHash|random(r).
e ProductData contains relevant information about production trading between the TDA and TDP, including product identi cation (PID), transaction timestamp (TxTS), transaction digest (TxHash), and a random number (r). e ProductData are generated in the process of product circulation. According to the relevant data, SKp and PKa are used to encrypt and sign PD, while ProductSign (PS) generated using formula (2) is used for authority authentication of TDA: ProductSign sign(sign(ProductData, SKp), PKa)|PID.
(2) e PS is then passed to the TDA, who can verify the product according to SKa and PKp. After PD is obtained, the TDA calculates the PD digest, simultaneously generates an authentication request, and then submits the request to the TDP to prove the traceability permission of the corresponding product. is completes the authentication process between the TDA and TDP. e authentication algorithm is shown in Algorithm 1.

Preprocessing Phase.
After the authentication phase, the system work ow enters the preprocessing phase. In this phase, the traceability purpose of the TDA is analyzed by the traceability feature generating algorithm. e core of the algorithm is based on public industrial production standards or key technical knowledge. According to these standards, knowledge, and related production data, traceability features can be generated.
As the production process progressed, data corresponding to the production process is generated in industrial production. Here, we de ne all the production processes as P, and we obtain the following equation: where P I represents one of the production steps in industrial production process, and, for any I, we use the following formula: D i Refers to the relevant data generated in a single production process; that is, there is a mapping relationship between the production process and the data, and each data point generated in the production process can be de ned by the following formula: In other words, the production process data can be de ned as a Boolean (bool), numeric (num), or hash type in a production record. Boolean data show the state of industrial devices, such as controllers or switches. Numeric data are used to record the data generated by the production devices like sensors. Hash data are used in industrial production to record relevant signatures. In this situation, a production process can be de ned by the following equation: where D I k is de ned as the kth data generated in the ith production process. erefore, for a single production product, its entire production process can be de ned by the following equation: On this basis, each industrial production process can be mapped to a set of data that holds all the data generated by  the production process. Considering the production data and the correlation of the traceability process, a function, dataParser, was introduced to filter the key data out of the set. e key data are then generated using the following equation: KeyData � dataParser(Product). (8) KeyData obtained through this process can be further processed and combined with relevant knowledge of the production field to generate traceability features. In this situation, the traceability purpose is defined as TP, and the traceability feature can be generated using the following equation: TraceabilityFeature � parse(KeyData, knowledge). (9) In the above equation, knowledge means the production knowledge that can be obtained by the relevant production experts or by a public production standard. Trace-abilityFeature is generated as a producer of zero-knowledge proof. On this basis, the TA will use TraceabilityFeature to generate a smart contract and also transfer the Trace-abilityFeature to TDP for generating the proof with raw data.
In addition to traceability features generated in the preprocessing phase, high-availability production data are provided by the TDP. Sensors generate raw production data during production, transportation, trading, and other activities, among which different production entities are distributed. ese data can be marked as rawData, and the rawData digest is marked as HashData. HashData and the identity of the data source form a transaction record, Tx. Finally, Tx is published to the data chain, which creates a relationship between production data and transactions. e on-chain data digest ensures the tamper-resistant and antirepudiation characteristics of the production data. e specific algorithm of data preprocessing is shown in Algorithm 2.
e data are first collected by a data collection device in the data physical layer and are used to extract the traceability features with production knowledge. e blockchain-based data record model has been previously discussed in the literature [34,37]. In our study, we only refer to those conclusions. e high-availability record of the production data provides a basis for the privacy protection approach to be adopted in the later construction phase.

Construction Phase.
After the two phases above are finished, the traceability features are provided with the available production data. e construction phase may import the zero-knowledge engine to generate related smart contracts and the proof. In the construction phase, the TA uses the traceability features generated by the preprocessing phase to create need-to-proof issues, which is called the witness, and then transfers it to the TDP. e TDP receives and generates the proof using a witness and the related raw data. After the proof is generated, the TDP submits the proof to the TA or the public, and then the TDA receives the data and verifies the proof.
At first, the generation of witness should be discussed. We define the witness as verifyKey, and three main issues should be proven. First, the available data must prove that the data that the TDP used to generate the proof are the same as those recorded in the data chain. Second, those data must be from the production being traced. ird, the traceability feature proof must show that the proof of data is satisfied After generating verifyKey, a zero-knowledge engine (ZKe) is introduced to the phase. e TA first uses ZKe to compile and set up with verifyKey to generate specific keys and transfer them to the TDP for creating a witness and proof. Simultaneously, the TA generates the smart contract using the genContract() function to provide an interaction interface for the traceability application layer. e TDP receives the keys to generating the witness using the related raw production data and then creates a proof that can be published to the public. e proof and contract generation algorithm is shown in Algorithm 4.
In the above algorithm, the TA publishes a smart contract address to the public to verify the proof. rough the TA, the TDP can first receive the traceability features and verification keys and then generate and publish the proof using raw production data. Meanwhile, the TDA can verify the proof through the smart contract in an open manner and finally reach traceability purpose in the verification phase.

Verification Phase.
After the completion of the above phases, the TA and TDP can provide the contract address, proof, and traceability features. e TDA can analyze the traceability features with common knowledge and judge the expected result. If the TDP fails to generate a related proof, the traceability failed, and the TDA and TA might negotiate new traceability features and restart the traceability process.
It was worth noting that traceability is based on traceability features that are generated with expert or public knowledge. Simultaneously, the related smart contract was permanently deployed in the blockchain, which means that the smart contract with the related proof could be reused to simplify the traceability process. e algorithm designed in the verification stage is shown in Algorithm 5.
In the verification phase, according to the judgement of traceability features, the TDA, TDP, and TA may complete the traceability process in an open and privacy-preserving way. e traceability system for this paper could be effectively applied to typical traceability scenarios such as anticounterfeiting verification [49], standard execution proofs [50], or abnormal investigations [51]. Furthermore, in the above relevant traceability process, the TA could be industrial entities or even production experts. ose TA may create a competitive environment for traceability feature generation. As more new technologies have been imported into the production process, a competitive relationship between TAs may be beneficial in raising the availability of the traceability system.

Analysis with Discussion
4.1. Security Analysis. Security issues in network attacks and privacy protection are discussed in this section. For network attacks, in the traditional traceability system, malicious intruders may launch Distributed Denial of Service (DDoS) attack or Advanced Persistent reat (APT) attacks in the centralized server [52]. Furthermore, the traceability data may be tampered with to destroy the traceability system's authority. However, the blockchain-based traceability system designed in this paper is immune to DDoS attacks and APTattacks [53]. All data digests stored in the blockchain for privacy protection are mapped with raw production data. With the importing of the zero-knowledge proof scheme, the traceability process can be transferred into a proof of related traceability features. e system described in this paper can avoid transferring the raw production data between any entities or individuals and so finally realize the traceability of privacy data. With the traceability system based on smart contracts, the entire traceability process could be monitored by the public, enhancing the traceability compliance and openness of the process. e system designed in this paper can be used to promote the industrial Internet, help it achieve a broader range of joint production, and provide an essential basis for trust [54].
Despite the network attack launched by malicious intruders, the blockchain itself may also have security issues such as transaction-related attacks (e.g., double spending attack), consensus failure, and smart contract-oriented attacks. In our work, we deployed the whole application on Ethereum environment to realize the availability of traceability process and assumed that the consensus process, smart contract execution, or other application process logic are operating without malicious behavior. Further research was deeply discussed in related works [55]; we just make the assumption above in order to reach the availability of traceability system.

Available Analysis.
In this paper, zkSNARK toolkit was introduced to form a privacy-preserving scheme. According to the generated traceability features, a smart contract could be published in the Ethereum blockchain. Our experimental environment included an i7-6700 CPU and 8 GB of memory; considering the highly available and widely used blockchain system, a widely used IDE with Ethereum testnet [56] was also used. We designed the test with 1 to 100 traceability features and generated a related proof and smart contract. However, the calculation of Ethereum is not free; the veri cation and the publication of a smart contract could require a computing fee (gas). In this situation, gas could be treated as a cost of traceability process, and the gas consumption of contract publication and veri cation is shown in Figure 3.
As shown in the gure above, the gas consumption does not signi cantly increase as the number of traceability features increases. e gas consumptions of smart contract publication and validation are approximately 1535000 and 242800. With the current price of the Ethereum, the costs are approximately 0.15 ETH and 0.02 ETH. As shown above, the publication and veri cation may increase the cost of traceability. In contrast, in traceability scenarios, the gas consumption could be optimized, such as through the production of a continuous linear process that could be implemented in a contract for multiple data validations and could nally reduce the gas consumption of publication and veri cation. At the same time, the system designed in this paper is built in the Ethereum testnet environment, and the traceability system in this paper can also be deployed into mainstream authorized blockchain systems, such as Hyperledger.
rough corresponding smart contracts, the extra cost of publication and veri cation may e ectively be avoided.

Comparative Analysis.
In the comparative analysis, this section compares our traceability system with the traceability systems constructed in related studies to demonstrate the advantages of our approach. In this paper, we choose three typical models of traceability systems: a traceability system designed without blockchain [11] (Centralized), a blockchain-based traceability system without a privacypreserving scheme [30][31][32][33][34][35], and a blockchain-based traceability system with a privacy-preserving scheme [36,37]. Although the traceability systems we choose to compare were designed for di erent purposes and industrial environments, the issues in each typical model are common; therefore, these traceability system models were selected for comparative analysis of data reliability, traceability target, privacy protection scheme, attack resistance, and cost of traceability. e analysis results are shown in Table 1.
Unlike other related traceability systems, this paper constructed a traceability system using a decentralized architecture based on blockchain; on this basis, unlike a centralized traceability system, data reliability and availability could be fully guaranteed. At the same time, our approach can defend against DDoS attacks, and, due to the privacy protection scheme of zero-knowledge proofs together with authentication, our approach realizes a traceability of privacy data in industrial production that is superior to other decentralized traceability systems and could e ectively build trust relationships between industrial entities. For the comparison of traceability cost, a centralized traceability system needs to spend much to build and maintain a traceability system, that is, the traceability system Security and Communication Networks manager may have the right to fix the price of traceability service. It may influence the cost of traceability system users.
In contrast, in our approach, pricing was determined by the calculation cost of the smart contract. e TA was introduced to make the price of traceability more flexible and open. e TDA and TDP would also benefit from the traceability process.

Conclusions and Future Work
With the development of industrial production, the industrial Internet may create many more opportunities for industrial entities of all sizes. As a key part of building industrial cooperation between entities, the traceability system plays a significant role in the development of industrial Internet. e traceability system designed in this paper achieves traceability for privacy production data, which makes it possible to objectively judge the quality of products in order to raise production quality or optimize supply chain structure. On the other hand, the privacypreserving scheme designed in this paper raises the willingness of parties to share data, which may take good effect to raise the production capacity or reduce the resource consumption.
In order to make a stronger trust relationship between industrial entities in industrial Internet, our work could be combined with anticounterfeiting system [57] to reach a higher data privacy level or applied in production lifecycle management to reach sustainable manufacturing [58] with privacy. We would research the feasibility of those application scenarios with our work in the future and our work could make contribution to meeting the privacy demand in industrial Internet to some extent.
It is worth mentioning that our work still needs to be improved. Firstly, in this paper, the traceability feature generation process still depends on traceability knowledge held by a third party or public production standard [59]. Future studies should focus on areas such as artificial intelligence or industrial big data. Traceability features will be an important research direction in our future work. Secondly, security issues also need to be further researched, especially the security-oriented in blockchain system and smart contract. Our work is based on the premise of nonmalicious blockchain nodes; security protection schemes should be further discussed in order to provide more secure environment for blockchain [12].
Data Availability e gas consumption data used to support the findings of this study are included in the article.

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