In addition to their economic value, multifunctional forests fulfill both ecological and social tasks. Thus, sustainable forestry impacts diferent areas. Although basic digitalization already exists in the forest-based sector and is being developed further, the integration of heterogeneous systems and the sharing of data between the stakeholders continues to be a challenge that requires mutual trust. To meet this demanding requirement, data trustees represent a possible approach. Thereby, the complexity of trustful data sharing is delegated to an intermediary that provides an easy entrance for data providers and consumers as well as software developers. The paper at hand presents the preliminary findings of an ongoing study on trustful data sharing in the forest-based sector in Germany. Based on a general outline of the concept of data trustees, this contribution examines various opportunities for the forest-based sector, identifies the associated challenges from a technical, user, and legal perspective, presents the use case of trustful sharing of harvester production data, and proposes an architecture to encounter the challenges while unlocking the opportunities.
of data trustees and data spaces and shows how they can of a data trustee is mentioned in [10]. However, in order be distinguished from each other. It also provides a short to meet the needs of the involved actors, the exact impleoverview of data trustees that are currently developed mentation can only be determined with direct reference or already in use. Finally, aspects of data handling in the to the field of application. For the healthcare sector, for forest-based sector are shown. example, Lauf et al. [8] propose four archetypes with different goals, "Data Brokerage Trustee", "Data Processing 2.1. What Is a Data Trustee? Trustee", "Data Aggregation Trustee", and "Data Custody Trustee".
In order to address the desire for trustful and sovereign In principle, though, it can be stated that a data trustee data sharing, especially between unknown actors, data should act neutrally. This does not mean that a trustee trustees have become increasingly important in recent cannot have any interests of its own, nor that any thirdyears. Yet, despite their basic promise to simplify the party interests must be ignored or disregarded. Instead, conditions of data sharing, they lack a clear definition [ 6, it means that a balance must be established between the 7]. Additionally, Lauf et al. [8] state that there are various interests of the diferent actors. Furthermore, every data other terms like "data cooperatives", "data stewardships", trustee should establish an environment of trust. This or "data brokers" whose definitions only partially overlap includes ensuring data security and transparency. Here, with concepts of data trustees and which yet are used the latter refers to both the use of data and the activities as synonyms. Consequently, it is essential to ensure a of the trustee itself [6, 7]. common understanding of the term "data trustee".
Reiberg et al. [6] show that the focus in literature 2.2. What Is a Data Space? is either on managing or sharing data. The former is closer to common definitions of the general term "trustee", The idea of data spaces emerged when it became apwhich sees trustees as entities that manage or control parent that a central data storage solution cannot solve property for other entities [9]. This is consistent with the problem of data handling any longer. Organizations what Lindner et al. [7] call the narrow understanding of had to handle ever increasing numbers of diverse data a data trustee, where they compare a data trustee to a sources [11]. It was thus not possible to physically inbank, which, instead of assets, manages data in a fiduciary tegrate all the data into a single data base, but instead relationship. On the other hand, in a wide understanding, chosen to leave it at the source and achieve integration data trustees are seen as an instance of trust in the sense on a semantic level. of a "trusted third party" concept. According to this, a Over the years, various initiatives have started to dedata trustee is an independent organization that ofers an velop data space concepts with diferent focuses. To cominfrastructure and, optionally, services to enable trustful bine forces, twelve partners, including IDSA and Gaia-X data sharing between data providers and data consumers. AISBL, set up and operate the European CommissionIn this case, the final decision on whether to share the funded Data Spaces Support Centre (DSSC) [12]. Their data is made by the actors themselves [7]. Consequently, goal is to establish data spaces in multiple sectors, while this wide understanding focuses more on sharing data enabling an interoperable data sharing environment. For than managing data. this, the DSSC investigates the needs of various data
Comparing these two perspectives, it is easier to imag- space initiatives, develops guidelines for common data ine a data trustee managing data without sharing it than spaces, like security requirements or cross-sector stanthinking of examples where a data trustee shares data dards for data sharing, and ofers support for the deploybut does not manage it. Therefore, it is possible to argue ment of data spaces [13]. In order to achieve a common that a definition with a focus on managing data is more understanding, the DSSC provides a glossary [14], in all-encompassing than only focusing on data sharing [6]. which a data space is defined as an “infrastructure that To combine both perspectives, data trustees can gener- enables data transactions between diferent data ecosysally be defined as "institutions that manage data or rights tem parties based on the governance framework of that to data on behalf of and in the interest of others. In the data space” (p. 5). course of their activities, trustees obtain control over data When setting up a data space, it should be considered and then use it immediately or at a later point in time that this topic is extensive. For example, the IDSA structo enable the data provider or third parties access to it" tures their RAM [5] into business, functional, process, (translated from [6], p. 7). This is also consistent with information and system layers, while Otto et al. [15] the authors’ understanding of a data trustee. identify a business-oriented and a legal perspective in
The design of a data trustee, especially when using addition to a technical one. The business-oriented defithis general definition, still leaves a lot of room for inter- nition focuses on the role of data spaces as a collaborapretation, e.g., regarding functional scope or technical tion format between diferent organizations. The legal components. A generalized starting point for the design perspective sees data spaces as intermediaries for data sharing comparable to a data trustee in a wide sense.
the National Access Point for mobility data in Germany.
Although it is labeled as a "market place" and not as a data trustee, there are some parallels. Transparent conditions as well as sharing data reliably and securely are aspects that are in the focus of the MDM but also part of a data trustee [20]. In addition, both the MDM and a data trustee can be central components of data spaces (cf. [21] and Section 2.3). This shows that regardless of the naming, there is a need for data trustee functionalities.
Section 2.1 and Section 2.2 show that there are various definitions and perspectives on data trustees and data spaces.
Thus, there are overlaps of varying degrees, which is why it is currently not possible to clearly separate data trustees and data spaces. Nevertheless, a diferentiation can be made based on the view of purposes. Data trustees 2.5. How Does the Forest-Based Sector are seen as institutions that manage data or the rights to Handle Data Today? data, while data spaces are seen as infrastructures that enable data transactions (cf. Section 2.1 and Section 2.2). The amount of data created in the forest-based sector
Thus, although it can be argued that data trustees form is huge. Examples of large data sources are inventory data ecosystems, it is not always possible to call them data of forest stands, machine data produced by forest a "data space", due to their stronger focus on managing machines, and environmental data produced by diferent data. Yet, data trustees can be part of a data space or even types of sensors. Data handling in the forest-based sector enable it in the first place, since they represent a compo- is closely connected to the use of geoinformation. The nent that enables trustful data sharing. From this, it can INSPIRE (Infrastructure for Spatial Information in the be concluded that although it is not mandatory for data European Community) directive (2007/2/EC) [22] of the trustees to build on the principles of data spaces, it is an European parliament establishes a legal framework for advantage with regard to future application possibilities the use of geoinformation in Europe. It basically obliges (e.g., being part of a data space, enable sharing with data the member states of the EU to provide free access to spaces) [6]. diferent types of geodata.
Although some open data standards exist, they are not 2.4. Existing Data Trustees applied consistently along the value chain, and only small parts of the data are standardized. Internationally, the Lindner et al. [7] show that, despite the existing de- most important standards are StanForD or StanForD2010 mand for such solutions, the realization of data trustees [23] for forest machine data and papiNet [24] for the in practice is still low. Nevertheless, there are already timber supply chain with a focus on the paper industry. data trustees in operation (in Germany eight in total), For data sharing in timber handling, there are a number of focusing on particular sectors. Other solutions are cur- national standards like FHPDAT in Austria, eFIDS [25] in rently under development, so that there will likely be Great Britain and ELDAT/ELDATsmart [26] in Germany. diferent realizations in the near future, for example, in For forest data and forest management national standards healthcare, aviation, maritime industry, manufacturing, are evolving, like Forestand [27] in Sweden, while in logistics, automotive and mobility [7]. Some concrete Germany, no such standard exists yet. examples are presented below. The relationship of the actors in the forest-based sup
Regarding the B2B sector, there are mainly customized ply chain varies greatly among countries and regions. solutions for business processes. Providers of such cen- Whereas in Germany and Austria the supply chain oftralized solutions include, for example, the German Bun- ten consists of many small independent companies, in desdruckerei with CenTrust [16], the Bochum-based com- Sweden and Finland larger companies usually control pany DATATRUSTEE [17] or Nortal (healthcare) [18]. A the whole supply chain from forests to the final wood cross-domain solution is the Data Intelligence Hub [19] product [3]. These diferences have a high impact on the provided by T-Systems International GmbH [7]. Building diferent mechanisms for data sharing. E.g., in Sweden, on data space technologies, T-Systems promises that it Biometria collects large parts of the forest machine data is possible to connect to any data space and exploit the and provides a central platform for data providers to acvalue of a provider’s data by sharing it on the provider’s cess their data. However, this model is more like a cloud terms and delivering it securely. As a pioneer of cross- storage for own data than a data trustee for data sharing. domain data sovereignty, the Data Intelligence Hub is The large Scandinavian machine manufacturers have currently in (early) regular operation [19]. The Mobility built up manufacturer ecosystems used by stakeholdData Marketplace (MDM) is a neutral exchange platform ers owning or using forest machines (e.g., John Deere that utilizes secure data and communication standards TimberManager [28], Komatsu Maxifleet [ 29], PONSSE (e.g., brokers for data exchange, prevalent internet proto- Opti [30]). Thus, sharing forest machine data is possible cols, certificates, signatures, process logging) and acts as inside these ecosystems, but it is dificult to integrate data from diferent manufacturers. Hartsch et al. [ 31] an- ation, service provider, and funding agency) to ensure alyze legal, social and economic requirements for using the shared data (activity reports, reports on expenditure, forest machine data to improve timber supply chains in invoices, funding decisions) is only used in the intended Germany. They identify legal issues and a missing tech- way. nical infrastructure as main obstacles for sharing forest Aside from targeted data sharing, a data trustee is machine data but agree on the benefits of data sharing particularly helpful to provide data to data consumers for the supply chains. previously unknown to the provider. This is especially
Digital solutions adopted in the forest-based sector true for scenarios where data needs to be aggregated, often apply to an isolated process [3]. Examples for such anonymized, pseudonymized or otherwise preprocessed isolated processes are described by Rönnqvist et al. [32]. to hide original data owners from data users. Thus, data sharing is usually either implemented in a pro- An exemplary use case for this scenario would be the prietary manner or uses outdated, insecure technologies provisioning of environmental data from sensors, e.g., such as attachments in emails. Attempts to find general on microclimate, soil condition, tree growth, sounds of solutions for open data sharing infrastructures are rare. the fauna etc. A data trustee could be used to aggregate Chen et al. [33] use open source software to develop an and analyze data from various sensor (network) operainfrastructure for communication in the forest-based sec- tors and ofer it to various interested parties, e.g., policy tor. Their work served as starting point for the presented makers ("What is the average tree growth rate and soil approach. condition in the Sauerland region?"), forestry contractors ("Are skid trails passable without damage to the soil?"), environmentalists ("What region has a high biodiversity 3. Opportunities based on sounds from the fauna?"), or dam operators ("What is the influence of the soil moisture of catchment In general, a data trustee can: enable new cooperation areas on the water reservoir?"). Examples for large instalbased on trustful sharing of highly sensitive business data, lations of such sensor networks are the TERENO network address concerns regarding legal certainty of data and [35] in Germany or the standard environmental monitorits usage, reduce transaction costs by providing a single ing program "ICP forests" [36, 37] in Europe and beyond. access point to several data sources, protect the data Data integration and data analysis are key in this kind of sovereignty of its users, counteract market distortions by scenario and could be provided by a data trustee. making previously locked-in data available, or simplify The case is similar for inventory data describing fordata-driven research activities [7]. est stands. Making this data available, especially on a
The concrete opportunities for a data trustee in the large scale, would help many stakeholders to make better, forest-based sector are manifold. The following exam- informed decisions and to optimize their planning and ples and applications are based on the authors’ experi- their processes [38]. A data trustee can provide the necence from many research projects in the forest-based essary, easily accessible environment to aggregate and sector and numerous discussions with stakeholders. In preprocess the heterogeneous data of large numbers of general, digital data and its collection is already ubiqui- forest owners (e.g., about 60 % of forested land in the EU tous in this domain. As one form of application, a data is privately owned [39]), while protecting their individtrustee can serve as the trustful intermediary for targeted ual business interests. Sharing this data, a forest owner sharing among immediate, mainly known partners along might be rewarded with better consulting, cheaper manthe value chain. Here, the main opportunities for data agement, public subsidies, and the good conscience to providers and consumers are the simplification of data support the green transformation. sharing in existing business relations as well as the op- As a final example, harvester production data holds a portunity to assert existing legal bindings on a technical high potential for the improvement of many processes level in terms of usage control. in forestry [40, 41, 42]. The standardized data format (cf.
An example case for this form of application deals with Subsection 2.5) makes it easy to handle and analyze. A the promotion of consulting services for private forest data trustee can be used for targeted data sharing along owners. To open the market to independent tenderers, the wood supply chain (forest owner placing the felling recent changes in German legislation made the previous order, contractor performing the felling order etc.) in a approach of many state forest authorities illegal to ofer trustful way, e.g., by trustfully handling sensitive busiconsulting services at a subsidized rate. In North-Rhine ness data like (GDPR-related) personal information of the Westphalia, this was encountered with a twelve-step pro- machine operator, the forest owner’s price matrix, or the cess [34] to apply for public funding that incorporates contractor’s exact performance and process data. Like extensive data sharing. In this scenario, a data trustee sensor or inventory data, production data is also interestcould provide the necessary trust among the business ing for third parties, as it provides valuable ground truth partners (forest owner and forest management associ- data directly from the forest. For example, wood buyers
Sawmill
tween data providers and consumers in existing business relations, and asserting existing legal bindings on a technical level in terms of usage control Ensuring trust between (previously unknown) actors and providing data to potential consumers by aggregating, anonymizing, pseudonymizing or pre-processing data (hiding sensitive data), e.g., sensor data aggregation, inventory data [38] and harvester production data sharing [46]
Service
Provider Forest
Forest Owners
Remote Sensing Data Simulation, Prognosis,
Control Wood
Pile Authorities
Forestry
Machines Environm.
Sensors Photos (from center top, right around):
A. Böhm, RIF; 4x Pixabay; Pixabay; A. Böhm, RIF; 5x Pixabay 4.1. Infrastructural Perspective like sawmills are interested in large-scale production data From the infrastructural perspective, general challenges to derive information about forest regions, e.g., in order for a data trustee lie in its ability to serve as a trustful to open up new markets. Harvester production data can data sharing infrastructure that guarantees secure and be used to predict inventory attributes [43], estimate tree reliable data sharing as well as suficient data quality, composition and volume [44], or model forest volume enables fair collaboration among participants, and main[45]. In this context, a data trustee can provide the neces- tains neutrality and transparency [7]. sary means to protect individual forest enterprises’ data In the context of the forest-based sector, data trustees and, thus, their business interests. The value of the data encounter additional challenges specific to this domain. is already discussed and quantified [ 46]. A data trustee The first challenge for a data trustee in the forest-based can also provide the necessary means to monetize its sector pertains to unreliable internet connectivity [47]. provision. Regularly, forest machines and individuals afiliated
Finally, a data trustee ofers the opportunity to allow within the forest-based sector engage in work activities a digital participation for all kinds of enterprises from within geographically isolated regions, where a permathe forest-based sector, especially for the large number nent internet connection is hard to guarantee [48]. Exof small enterprises, by taking care of the challenges (see pecting the return of forest machines and subsequent next section) of trustful data sharing. A short overview data transmission inevitably leads to information delays of the presented opportunities is shown in Table 1. and thus potential loss of benefits.
Secondly, the forest is a transient environment [49].
In contrast to environments like the factory floor, the 4. Challenges conditions in the forest are constantly changing. Even with a reliable internet connection, this dynamic enviCompared to other sectors such as supply chain or the au- ronment presents a challenge in facilitating real-time tomotive sector, a data trustee in the forest-based sector is data sharing. For instance, dense trees and vegetation confronted with general and domain-specific challenges, in the forest may lead to signal attenuation or blockoften rooted in the sector’s heterogeneity (Figure 1). For age, wildlife may touch or damage sensors or devices, a data trustee in the forest-based sector, three main per- and severe weather conditions, including heavy rain or spectives are taken into account: infrastructural perspec- strong winds, may interrupt signal transmission, and tive (Section 4.1), user perspective (Section 4.2) and legal spontaneous incidents such as fire accidents can cause perspective (Section 4.3). To summarize this chapter, a extensive damage to equipment and further disrupt the short overview of challenges in the forest-based sector is signal transmission [50]. For a data trustee, therefore, shown in Table 2. developing strategies to address this ever-changing forest environment becomes imperative. trustee in the forest-based sector to efectively manage
Moreover, a distinguishing feature in the forest-based and coordinate data sharing processes. sector is its multitude of independent and heterogeneous systems (see Figure 1) [47]. As previously discussed in 4.2. User Perspective Section 3, data trustees provide the ability for various forest actors to aggregate and preprocess the large amount From a user perspective, general challenges of a data of heterogeneous data. However, this presents an in- trustee are whether data providing users are willing to evitable challenge for a data trustee as well. Stakeholders delegate usage control to a third party, whether data and actors employ specialized machines, devices and soft- consuming users can rely on quality and legal certainty ware applications or services in diferent heterogeneous of provided data, or whether the actual added value of systems. In the past few years, a diverse array of dig- the data trustee can be clarified to ensure its day-to-day ital technologies, including RFID, GPS-based tracking operation on a permanent basis and to expand it as far devices, and light detection and ranging (LIDAR), have as possible [7]. been efectively utilized to gather data in the forest-based Focusing on potential users in the forest-based sector, sector [3]. This data generated from heterogeneous sys- there are specific challenges to overcome for the impletems often exhibits non-uniform data formats, which mentation of a data trustee, as well. raises several issues that data trustees need to address, Many people employed in forestry are older (e.g., in such as how to eficiently collect diverse data sets from 2019, more than a quarter were aged above 50 years [39]). various systems and make most of them usable, how to Likewise, private forest owners are older (e.g., a study preprocess and analyze data sets with non-harmonized from 2010 showed a large proportion to be older than formats, as well as how to aggregate and integrate this 60 years [52]). Although older adults more and more data and deal with data compatibility issues. overcome the digital divide, they still lag behind [53].
Additionally, as elucidated in Section 3, the involve- Besides age, there’s a digital inequality in terms of the size ment of a data trustee provides a valuable opportunity of companies. An estimated 80-90 % of forest enterprises for participants of various enterprise sizes. Especially for are Micro-, Small- and Medium-sized Enterprises (MSME) small- and medium-sized enterprises, their interests lie in [54]. Due to limited resources, they often do not have using a data trustee for data sharing and analysis, thereby extensive IT knowledge or interest in dealing with legal expanding their business and enhancing profitability in a framework conditions. cost-efective manner. In contrast, however, these enter- On the other hand, stakeholders in the forest-based prises are disinclined to invest excessive cost and efort sector - especially in Germany - have major concerns in IT infrastructure, and they often lack deeper IT pro- about sharing information with others. The authors’ ifciency. Consequently, from a technical perspective, a many years of experience from discussions with stakecritical requirement for a data trustee in the forest-based holders revealed a wide variety of reasons. Contractors sector is to facilitate a streamlined and accessible process fear to reveal information that allows to derive exact perfor these potential users to seamlessly join and leverage formance figures from production data that might give the infrastructure. clients an advantage in price negotiations. They also fear
Last but not least, the forest-based sector poses a chal- intense surveillance regarding adherence to nature conlenge for a data trustee in terms of organizing data shar- servation guidelines (harvesting measures at the wrong ing processes involving multiple actors, while ensuring time or place). Forest owners want to protect their crititransparency and traceability. In Germany, for example, cal business data when negotiating with contractors (e.g., the harvesting of timber is typically carried out by con- price matrix) or buyers like sawmills (e.g., exact locations tractors on behalf of the forest owners. The contractor- of preferred wood qualities). Besides, there are conflicts owned harvesters collect a wealth of data on parame- of interest towards environmentalists regarding wood ters such as length, thickness, and quality of the individ- usage versus environmental protection (land set-aside) ual logs during felling and subsequent processing [51]. [4] as well as a general fear of intensive audits by state This scenario exemplifies the complexity faced by data authorities. trustees: when one of the users of the data trustee ex- The key requirements for a data sharing solution in presses interest in harvester production data, the forest the forest-based sector are therefore ease of access, ease owner, contractor and harvester all become associated of use, and trustworthiness. All these requirements can with this data set. The involvement of multiple actors be met by a data trustee. in a single data sharing process significantly increases the complexity for a data trustee, as it needs to organize 4.3. Legal Perspective the workflow in a proper manner and navigate transparency as well as traceability requirements at diferent The data trustee will handle both personal data and malevels. This presents a crucial requirement for a data chine generated data.
In the EU and the European Economic Area (EEA), ation services and will significantly influence the design handling personal data is strictly regulated by the Gen- and configuration of data trustees within the EU and eral Data Protection Regulation (GDPR). This regulation EEA in the following years. It remains to be seen if the is part of the EU privacy law and human rights law and EU DGA will promote the acceptance of data sharing focuses on data protection and privacy. It was an inspi- solutions or if its strict regulations for providing a data ration for several other laws concerning the protection sharing solution will slow down the development of new of personal data. For instance, the California Consumer data trustees [7].
Privacy Act (CCPA) possesses many similarities with the
The core element of the GDPR is to enable individuals Overview of challenges in forest-based sector to determine in which way their personal data may be saved, processed or transferred to third parties by data Challenges processors. Individuals – data subjects – possess full Infrastructure Unreliable internet connectivity [48] ownership of their personal data. The GDPR provides perspective The forest is an ever-changing environa very strict framework concerning, inter alia, rights of ment [49] data subjects, duties of data controllers or processors, Numerous independent heterogeneous and transfers of personal data to third countries. Once systems and non-uniform data formats a data trustee has to process personal data, obeying the [47] GDPR is mandatory. vPernovieidnitnjgoiunsienrgs awnidthuasasgi mepplreocaensds con
From the legal perspective, the major challenge for Organizing data sharing processes ina data trustee will be handling machine generated data. volving multiple actors while ensuring In contrast to the strict regulations of the GDPR, the transparency and traceability German legislator does not provide any concept of own- User Many forestry practitioners and private ership of machine generated data. This lack of regulation perspective forest owners are older [39] [52] causes massive uncertainty and confusion of anybody Requirements for low technological who is interested in sharing, processing and transfer- barriers and simplified onboarding proring machine generated data. Data sharing solutions in cess particular sufer from the nonexistent legal definition. Stakeholders have major concerns The stakeholders’ willingness to ofer machine generated about sharing information with others data is very low since they suspect the loss of sovereignty Legal Ldaucektoofvoawrinoeursshreipascoonnscept for machineof data their machines generated (e.g., losing company perspective generated data secrets, legal implications). As mentioned in the previous Mandatory definition of ownershipsection, this is a crucial problem in implementing a data like claim for machine-generated data sharing solution in the German forest-based sector. [31]
Consequently, a data trustee that respects everyone’s Ensuring technical enforcement of the data sovereignty and enables adjusting usage rights for established usage rights transparently the data entrusted to it seems to be a possible solution Compliance with EU Data Governance for data sharing in the German forest-based sector. This Act (EU DGA) [7] causes two challenges.
Firstly, a definition of an ownership-like claim for machine generated data has to be defined and has to be mandatory within the data ecosystem of the data trustee. 5. Use Case One possibility to establish this ownership-like claim is given by the General Terms and Conditions (GTC) that As mentioned in Section 2.1, developing a data trustee every participant in the data ecosystem of the data trustee requires a reference to the field of application. For this, has to accept [31]. Moreover, a catalog of usage rights the authors consider the use case "trustful sharing of templates needs to be developed and provided for the harvester production data". Its implementation with a stakeholders. data trustee is shown in Figure 2. It consists of five actors:
Secondly, the data trustee should be able to transpar- • A Forest Owner1 who wants to sell wood after ently ensure the technical enforcement of the established trees are felled and is therefore interested in prousage rights as far as technically possible. This might be duction data. He is also willing to share data for the major challenge. research purposes.
Another legal challenge a data trustee has to deal with is the recently enacted EU Data Governance Act (EU 1from here on, terms in italics refer to either elements of Figure 2 or DGA). It defines conditions for providing data intermedi- Figure 3 • A Contractor who is hired by the Forest Owner to agreement, the actors can use certain parts of the HPR in fell trees. diferent ways. For example, the Forest Owner is not able • A Harvester and its operator. The machine is to see the personal data of the harvester operator, while owned by the Contractor and used to fell trees. the Contractor is not able to see the tree positions. All While felling trees, harvester production data other actors who agree to the "general ofer" obtain iden(HPR) is generated. tical rights to the HPR. Only the agreements difer in the • A Sawmill that wants to buy wood and is there- respective assignee, e.g., Sawmill or Research Institute. fore interested in HPR to check what stem sizes The focus of the initial development is supposed to be are available. on trustful data sharing. The complexity is significantly • A Research Institute that investigates forests in increased if several actors can edit a data ofer. For this a certain area and wants to update forest stand reason, the first step assumes that UR are coordinated data by using HPR. outside the Data Trustee, so that an ofer is created on behalf of all actors who have rights to the data. However, The machine generated HPR conforms to the Stan- the integration of the collaborative creation of a data ForD2010 standard mentioned in Section 2.5 and con- ofer into the Data Trustee is of crucial importance for tains information that relates to diferent actors or their both usability (e.g., tool for creating machine-readable property. Consequently, diferent actors have rights to UR) and security (e.g., ensuring that one’s own UR are the machine generated data. set exactly in the desired way) and should therefore be
What is the advantage to use a data trustee in this considered in the near future. use case? The machine-generated data contains information that is on the one hand valuable for the diferent UHRPR == HUsaarvgeestreigrhptsroduction data actors and is on the other hand critical to share, because AO == AOgffreerement some of the actors are either not allowed to share the SP = Starting point Contractor data with other actors (due to the GDPR), or do not want Forest Owner (SP) ttioonshianrcel uthdiessinthfoerimdeantitoitny., Ewxoarmkpinleg ohfocurrist,icaanldinpfoerrmfoar-- JUoRb JUUoRRb Harvester mance of the harvester operators, which can only be HPR A shared with the explicit consent of the harvester oper- HPR A HPR ator according to the GDPR. The forest owner can use O O O the machine-generated data for operational control and Data to take account of how much timber is actually removed HPR Trustee from his forest and how much timber should still be left. O O O The Sawmill can use the data to improve the accuracy of the delivery forecasting and for organizing the timber HPR A HPR A logistics. The Research Institute might use the data to account for the harvested timber. Finally, the Contractor Sawmill Research Institute can be relieved of manually handling the data sharing, Photos: Pixabay as it is often the case today.
The red dashed lines in Figure 2 show the commu- Figure 2: Data Trustee use case “trustful sharing of harvester nication outside the Data Trustee. In the first step, the production data”.
Forest Owner sends a Job to the Contractor together with what he considers to be valid usage rights (UR) on the HPR, e.g., "filter out tree coordinates". The Contractor adds UR from his perspective, e.g., "filter out personal 6. Architecture data of the harvester operator", and sends the package to the respective Harvester. Based on the UR defined by As shown in Section 4, a data trustee for the forest-based Forest Owner and Contractor, diferent ofers ( O) are cre- sector needs to address specific challenges that might not ated, one addressing the Forest Owner (left O, light blue), be present in other industries like an unreliable internet one addressing the Contractor (middle O, dark blue) and connection or a very heterogeneous user and developer a general one, addressing a general audience (right O, base, especially in terms of IT infrastructure, IT expertise, green). After finishing the Job, the Harvester automati- or legal expertise. As far as known to the authors, there cally combines the three ofers with the HPR and sends is no existing data trustee that already fulfills all these the resulting package to the Data Trustee. Subsequently, aspects. Therefore, this paper proposes a preliminary other actors can view the respective ofers. If an ofer is architectural design regarding the use case mentioned accepted, an agreement (A) is reached. Depending on the in Section 5. The intention is to leverage the opportunities and tackle the challenges presented in Section 3 and Section 4 centering the architecture around the Data Trustee and making use of various components proposed by the IDSA in their RAM [5]. The following sections introduce requirements to support trusted data sharing at the technical and ecosystem levels (Section 6.1), the Data Trustee’s architecture (Section 6.2) and outline the ongoing and planned implementation of it and the infrastructure around it (Section 6.3).
At the technical level, relevant requirements can be derived from the aforementioned opportunities and challenges (cf. Section 3 and Section 4). The designed architecture needs to support reliable data sharing, provide low technical threshold for user onboarding and usage, and maintain transparency and traceability even in complex scenarios with multiple users. In addition, existing legal bindings must be enforced at a technical level with approaches of usage control.
From the ecosystem perspective, the most important aspect is creating trust, otherwise stakeholders will not be willing to share data. Consequently, an important requirement is providing a trustful environment that is built upon a foundation of mutual trust. Further requirements from the user and legal perspective can be derived from Table 2 but will not be presented in detail here. mediary: data sources (on the top in Figure 3) provide data and associated rights, which are then handled by the Data Trustee (middle) and finally obtained by data 6.2. Components and Data Flow users (bottom).
The relationship between data trustees and data spaces is Based on the concepts of the IDSA, where connectors the primary focus of Section 2.3. This section concludes represent the major gateway for communication in data that data trustees act as institutions managing data and spaces and have the ability to perform usage control as a their respective rights, while data spaces provide the policy enforcement point [55], the central component of infrastructure that facilitates data transactions. It fur- the proposed Data Trustee is the Data Trustee’s Connector. ther proposes that data trustees can be built efectively Even though the use case presented in Section 5 fousing data spaces as a foundation because data space cuses on machine-generated HPR and the Harvester as components as defined by IDSA can carry out many of the data source, the authors’ goal is to consider transthe expected functions of a data trustee. This approach actions that can be both manually initiated by human potentially enables data sharing between a data trustee users and automatically triggered by machines. This reand data spaces. The first conceptual step to implement sults in more diverse application possibilities, while at this approach is the identification of necessary data space the same time facilitating a system rollout in practice. components for a minimum viable data trustee. Following this, the Data Trustee’s proposed architecture
To this end, Figure 3 shows an overview of the pro- allows diferent types of data sources and users: posed Data Trustee’s architecture with a focus on the data flow between important components. A part of the • Human users that require a User Interface (Data ongoing study will be to show, on the basis of an im- Owner with Data Source 1 and Data User 1), e.g., plementation, which data space components are indeed Forest Owner or Contractor as an individual with minimal IT knowledge, necessary for a minimal viable data trustee, and which are optional.
The basic path of data flow for a Usage Agreementbased data transaction2 uses the Data Trustee as an interthat outlines the terms of data use. This agreement is established either through active negotiation between the data provider and consumer, or by a consumer discovering the data ofer and agreeing to the provider’s pre-set terms. The authors refer to this process as a "Usage Agreement-based data transaction". 2In the context of data spaces, data transactions typically follow a process of negotiation or discovery, resulting in a Usage Agreement • Human users that do not require a User Interface components that are particularly interesting for the reor machines without data space connectors (Data alization of the opportunities in the forest-based sector Source 2 and Data User 2), e.g., Harvester that au- as described in Section 3. An important component to tomatically uploads HPR after executing a job or ensure the availability of ofered data is Data Storage, as Sawmill using an own software to automatically many potential participants of a common forest-based analyze HPR, and data sharing infrastructure are not able to host data on• Actors with own connectors (Data Source 3 with premise and guarantee continuous access to it. This can its providing connector Connector IN and Data be due to missing IT knowledge, an unreliable internet User 3 with its consuming connector Connector connection (cf. Section 4.1) or sources like the Harvester OUT —named from the trustee’s perspective), e.g., being turned of after usage. Furthermore, many highly Research Institute with significant IT knowledge. beneficial use cases are only possible by removing or modifying sensitive data (e.g., Harvester operator names or exWhile the latter type can communicate via the Connector act tree positions in HPR, cf. Section 4.2), achieved using directly, the former two require a Data Service Interface, techniques like filtering, pseudonymization, anonymizaas a part of the Data Trustee, that supports protocols to tion, or aggregation. Services to realize these functions directly communicate with the respective actor or can or perform more domain-specific calculations like sensor provide a backend for the User Interface. This, in turn, data processing for wood harvest yield forecasting, cliforms the Data Trustee’s frontend to the human user, pro- mate impact analysis, or business process optimization viding an interface to create, modify and delete data ofers need to be accessible via the Data Trustee, making its (including the definition of rights), negotiate agreements support for Apps3 sensible. In the given use case, this and to receive or view data. These two types provide allows the Research Institute to not only analyze data but seamless opportunities for potential participants to join provide their analysis algorithms, e.g., determination of and utilize the Data Trustee. They ofer a user-friendly wood harvest eficiency, as an App, e.g., to the Contractor approach, eliminating the need for extensive IT expertise, for business optimization processes. While both, Data enabling easy integration with the Data Trustee, secur- Storage and Apps, can be considered parts of the Data ing data sharing and maintaining the data sovereignty, Trustee itself, the Data Trustee should at least provide particularly for machine-generated data (cf. Section 4.3). interfaces to existing services with these functionalities. The User Interface also acts as a singular marketplace In the sense of decentralization and data sovereignty, this for the envisioned application of a common trusted and allows the user to choose where data is stored and whose sovereign data sharing ecosystem. Nevertheless, the vi- apps are used. In addition, the integration of external sion of data federation suggests that data trustee func- apps leads to an increased number of potential uses. tionalities might as well be distributed. In this case, the infrastructure needs to be designed for multiple trustees.
This introduces additional complexity and challenges, 6.3. Implementation Approach particularly, around ensuring interoperability and man- The proposed Data Trustee is based on the concepts proaging the coordination and cooperation among diferent vided by the IDSA, as the initiative’s technology is matrustees, e.g., handling usage policies. While this will be ture, and their components are already available for seconsidered as a perspective during implementation, the cure data sharing. While Gaia-X represents a broader focus will lie on a single, scalable data trustee. As men- and more modern approach to federated data and sertioned before, the proposed Connector forms the center vice infrastructure, its relative immaturity means that of the Data Trustee. Nevertheless, the final implementa- IDSA currently provides a more practical foundation for tion might require the inclusion of additional connectors a data trustee. Importantly, the current choice does not for providers and consumers who do not have an own compromise future interoperability with Gaia-X, as both connector. The User Interface’s backend is a potential initiatives are member of the DSSC and committed to location for these. compatibility.
The last mandatory component of the Data Trustee is Connector, Logging, and Apps are functional blocks of the Logging functionality. With it, all communication, the proposed Data Trustee that can be realized using Inaccess to and technically traceable usage of data is stored ternational Data Spaces (IDS)4 components—connectors, in the Data Trustee to provide transparency and enable clearing houses, and apps, respectively. Many implemenaccountability, thus, opening the possibility for the mon- tations exist, such as the Eclipse Dataspace Connector etization of data provision. This allows Forest Owner and [56] and the IDS Clearing House prototype implemenContractor to track how the data to which they hold rights has been used and, if desired, to generate an invoice for the use of the data.
In addition, the proposed architecture features optional 3In the IDSA sense of code-to-data 4IDSA and IDS are related as IDS is the concept for a secure and trusted data sharing environment, while IDSA is the association that works to develop, implement and promote the IDS concept [5]. tation [57]. IDS Apps, due to their domain specificity, tion is still restricted due to limited communication and are components that have to be developed individually data sharing. The opportunities for the sector by using a and can be distributed using IDS App Stores (see [58] for data trustee seem manifold—from sharing environmental, an implementation) after a defined certification process. process and production data between immediate busiThe usage policies to be enforced by the connectors are ness partners to optimizing the supply chain to providbased on the Open Digital Rights Language (ODRL)5. ing added value through secondary use by third parties
As the proposed architecture focuses on the develop- (scientists, policy makers, environmentalists, dam operment of an IDS-compliant data trustee, it does not feature ators ...), data refinement and monetization. However, all components necessary to establish a data space but the identified challenges, ranging from infrastructural is designed to allow seamless integration into them and to user-related to legal issues, cannot be ignored. The relies on further components for its proper functioning. authors assume the proposed architecture to be a viable Obligatory components include, in particular, an iden- approach to resolve these challenges, while unlocking tity provider to enable authentication and (user- or role- the outlined opportunities for the forest-based sector and dependent) authorization for data usage. Additionally, enable the evaluation of viable business solutions for a a metadata broker is required to facilitate querying the trustful data sharing ecosystem in a domain that is highly metadata of services (including, e.g., filtering, anonymiza- diverse with respect to roles and the economical as well tion, and value-added analysis services) and data ofers 6. as ecological scale.
Finally, vocabularies serve to ensure a common under- In the context of their ongoing study, the authors will standing of the various terms used to describe the data focus on the presented use case of harvester production and services provided. This facilitates the analysis and data and its sharing between forest owner, contractor, integration of data in diferent formats. harvester, sawmill and research institute. Their next steps
For the findability of relevant data by human users, an- will be the prototypical implementation of the architecother important component is some kind of data market- ture to assess its suitability using practical examples from place (not a defined component of the IDS infrastructure). this use case. This prototype, within a realistic testing From the frontend perspective, as mentioned before, this environment, will help to understand the practicability of relies on the User Interface that is already part of the the presented architecture with respect to requirements Data Trustee. The corresponding backend relies on the and expectations. metadata broker which allows appropriate searches for The authors are aware that there are other issues rel(meta)data, and, potentially, on vocabularies to prevent evant for the future with regard to data rights that are misunderstandings by ofering user-specific data and ser- not addressed in this paper. Due to the focus of the ongovice descriptions. This can be particularly useful in the ing study on machine-generated data, data generated by forestry-based sector, as the terms employed there can other techniques, like machine learning models, is not vary depending on the region and the role of a given discussed within the scope of this study. Identifying the participant of the envisioned data sharing ecosystem en- diferences involved could be the focus of a future study. abled by the Data Trustee (cf. Section 4.1). The authors’ approach is to provide the User Interface as a web application, as this does not require the user to install any Acknowledgments specific software and allows access from diferent devices, like computers, smartphones or tablets. Furthermore, it can be assumed that users with little IT knowledge, e.g., Forest Owner of advanced age (cf. Section 4.2), are more likely to use a web application, if it is self-explanatory, instead of performing a possibly complicated software installation. Consequently, this implementation is intended to keep the entrance barrier for the user low.
This work was supported by the Federal Ministry for Education and Research (BMBF), Germany [grant number 16DTM102A..D], and funded by the European Union - NextGenerationEU. The views and opinions expressed are solely those of the authors and do not necessarily reflect the views of the European Union or the European Commission. Neither the European Union nor the European Commission can be held responsible for them.
The forest-based sector plays an important role in the green transformation of our economy. Yet, its digitaliza5ODRL is a policy expression language and an endorsed W3C Recommendation since 2018 [59]. 6Identity provider and metadata broker are not featured in Figure 3 due to its focus on data flow.
Funded by the European Union
NextGenerationEU http://dx.doi.org/10.3390/f12040460. doi:10.3390/ vian Journal of Forest Research 36 (2021) 306–314. f12040460. doi:10.1080/02827581.2021.1919751. [32] M. Rönnqvist, S. D’Amours, A. Weintraub, A. Jofre, [43] A. Saukkola, T. Melkas, K. Riekki, S. Sirparanta, E. Gunn, R. G. Haight, D. Martell, A. T. Murray, J. Peuhkurinen, M. Holopainen, J. Hyyppä, M. VasC. Romero, Operations research challenges in taranta, Predicting forest inventory attributes usforestry: 33 open problems, Annals of Opera- ing airborne laser scanning, aerial imagery, and tions Research 232 (2015) 11–40. doi:10.1007/ harvester data, Remote Sensing 11 (2019) 797. s10479-015-1907-4. doi:10.3390/rs11070797. [33] J. Chen, M. Schluse, H.-J. Roßmann, Enabling a [44] J. Rasinmäki, T. Melkas, A method for estimating Secured Communication in Distributed IoT Using tree composition and volume using harvester data, the Smart Systems Service Infrastructure, in: 2021 Scandinavian Journal of Forest Research 20 (2005) IEEE International Conference on Pervasive Com- 85–95. doi:10.1080/02827580510008185. puting and Communications Workshops and other [45] M. Hauglin, E. Hansen, E. Sørngård, E. Naesset, Afiliated Events (PerCom Workshops), SPT-IoT, 5. T. Gobakken, Utilizing accurately positioned harIEEE PERCOM Workshop on Security Privacy and vester data: modelling forest volume with airTrust in the Internet of Things, Kassel (Germany), borne laser scanning, Canadian Journal of For22 Mar 2021 - 26 Mar 2021, IEEE, 2021, pp. 674–679. est Research 48 (2018) 913–922. doi:10.1139/ URL: https://publications.rwth-aachen.de/record/ cjfr-2017-0467. 818947. doi:10.1109/PerComWorkshops51409. [46] A. Kaulen, D. Mayer, Runder Tisch "Datenschatz 2021.9431124. Forstmaschine", FTI - Forsttechnische Informatio[34] UNIQUE forestry and land use GmbH, Waldbauern- nen (2022) 18–19. Available in German only. lotse, 2023. URL: https://www.waldbauernlotse. [47] R. Sahal, S. H. Alsamhi, J. G. Breslin, M. I. Ali, Innrw/, available in German only. dustry 4.0 towards forestry 4.0: Fire detection use [35] H. R. Bogena, TERENO: German network of ter- case, Sensors 21 (2021) 694. URL: https://doi.org/10. restrial environmental observatories, Journal of 3390/s21030694. large-scale research facilities JLSRF 2 (2016). doi:10. [48] R. Singh, A. Gehlot, S. V. Akram, A. K. Thakur, 17815/jlsrf-2-98. D. Buddhi, P. K. Das, Forest 4.0: Digitalization [36] M. Lorenz, International co-operative programme of forest using the internet of things (iot), Jouron assessment and monitoring of air pollution ef- nal of King Saud University-Computer and Inforfects on forests-icp forests, Water, Air, and Soil mation Sciences 34 (2022) 5587–5601. URL: https: Pollution 85 (1995) 1221–1226. //doi.org/10.1016/j.jksuci.2021.02.009. [37] T. I. of Forest Ecosystems, Icp forests, 2023. URL: [49] J. Reitz, M. Schluse, J. Roßmann, Inhttp://icp-forests.net/. dustry 4.0 beyond the factory: An ap[38] A. S. Kangas, Value of forest information, Euro- plication to forestry (2019) 107–116. pean Journal of Forest Research 129 (2009) 863–874. doi:10.1007/978-3-662-59317-2_11. doi:10.1007/s10342-009-0281-7. [50] A. Salam, Internet of things for sustainable forestry, [39] European Commission. Statistical Ofice of the Internet of Things for sustainable community deEuropean Union., Agriculture, forestry and fish- velopment: Wireless communications, sensing, and ery statistics: 2020 edition., Publications Ofice, systems (2020) 147–181. URL: https://doi.org/10. 2020. URL: https://ec.europa.eu/eurostat/en/web/ 1007/978-3-030-35291-2_5. products-statistical-books/-/ks-fk-20-001. doi:10. [51] T. Purfürst, J. Erler, The human influence on pro2785/143455. ductivity in harvester operations, International [40] M. Maltamo, M. Hauglin, E. Naesset, T. Gobakken, Journal of Forest Engineering 22 (2011) 15–22. URL: Estimating stand level stem diameter distribution https://doi.org/10.1080/14942119.2011.10702606. utilizing harvester data and airborne laser scanning, [52] F. Schmithüsen, F. Hirsch, Private forest ownSilva Fennica 53 (2019). doi:10.14214/sf.10075. ership in europe (2010) vi, 110 p. :. URL: http:// [41] J. Kemmerer, E. R. Labelle, Using harvester data digitallibrary.un.org/record/697427, at head of title: from on-board computers: a review of key findings, United Nations Economic Commission for Europe/opportunities and challenges, European Journal Food and Agriculture Organization of the United of Forest Research 140 (2020) 1–17. doi:10.1007/ Nations, Forestry and Timber Section, Geneva, s10342-020-01313-4. Switzerland. [42] J. Söderberg, J. Wallerman, A. Almäng, J. J. Möller, [53] J. Francis, C. Ball, T. Kadylak, S. R. Cotten, Aging in E. Willén, Operational prediction of forest at- the digital age: Conceptualizing technology adoptributes using standardised harvester data and air- tion and digital inequalities, in: Ageing and Digital borne laser scanning data in sweden, Scandina- Technology, Springer Singapore, 2019, pp. 35–49. doi:10.1007/978-981-13-3693-5_3. [54] E. Pohnan, B. Cammaert, T. Cavanagh, Enabling micro, small and medium-sized enterprises to participate in legal timber production and trade – Transformational changes generated by the FAO-EU FLEGT Programme, number 189 in FAO Forestry Paper, FAO, Rome, Italy, 2022. doi:10.
4060/cc3107en. [55] L. Zhi, W. Jing, C. Xiao-su, J. Lian-xing, Research on policy-based access control model, in: 2009 International Conference on Networks Security, Wireless Communications and Trusted Computing, volume 2, IEEE, 2009, pp. 164–167. doi:10.1109/
NSWCTC.2009.313. [56] E. Foundation, Eclipse dataspace components, 2023. URL: https://projects.eclipse.org/projects/ technology.edc. [57] F. AISEC, Ids clearing house github repository, 2023. URL: https://github.com/Fraunhofer-AISEC/ ids-clearing-house-service. [58] F. FIT, Ids appstore github repository, 2023. URL: https://github.com/ International-Data-Spaces-Association/
IDS-AppStore. [59] S. V. Renato Iannella, Odrl information model 2.2, 2018. URL: https://www.w3.org/TR/odrl-model/.