Blockchain: Implications For Photogrammetry.

I found this article from STUART I. GRANSHAW in ´The Photogrammetric Record´ while doing some current research and thought I´d share it in a blog format for a change.

The Photogrammetric Record © 2021 The Remote Sensing and Photogrammetry Society and John Wiley & Sons Ltd

From 2021 but still well worth the read. Enjoy

SEVERAL RECENT EDITORIALS have touched upon open data, big data, and the fourth industrial revolution. For example, Granshaw (2018) outlined big data and the dominant role played by global tech giants such as Alibaba, Amazon, Apple, Baidu, Facebook, Google, Microsoft, Samsung, and Tencent. This was complemented by Granshaw (2019) which concentrated on open data, promoting free access not only to the results of research but also to the underlying data itself. Both these editorials outlined the fourth paradigm, proposed by Jim Gray, where data-intensive science is integral to modern research and is reflected in other “ fourth ” initiatives, such as Industry 4.0.

Blockchain is a word that is new to me and perhaps many readers. Dowman (2018) provided a glimpse of this technology in a geospatial context, based on Ellehauge (2016).

Many applications seem peripheral to photogrammetry, yet this technology is becoming more widespread and its implications for our discipline, including big data and open data, make it worthy of further investigation. In the next two sections, I will attempt to provide some background to this technology: if this appears too dry and technical for an Editorial jump to the subsequent sections which provide concrete examples and indicate blockchain potential for photogrammetry and geomatics in general.

Whilst the term blockchain may be unfamiliar, it forms the technical basis of Bitcoin and other cryptocurrencies. With origins back to 1982, Kwong (2019) notes further work by two researchers, Stuart Haber and W. Scott Stornetta in 1991, who wanted to improve the security of files stored on personal computers by having dispersed, but interconnected, copies of a shared ledger with digital timestamps. The primary development emerged when the person (or group) under the pseudonym Satoshi Nakamoto (whose real identity is still unknown) implemented blockchain as the core technology of bitcoins. Rathore et al. (2020) report:

Bitcoins were developed after the financial crisis of 2009 as an alternative to traditional currency. It is widely considered that one of the many reasons behind this crisis was the single point of failure, exemplified by how centralized banks maintained financial records. There was no oversight in this process, hence the lack of fault-tolerant checks and balances. . . . (Blockchain) is a type of mechanism that validates, verifies, and confirms the transactions by recording them in a distributed ledger of blocks. It implements a consensus protocol to arrive at an agreement on the validity of a transaction by creating a chain of blocks.

This immutable chain of blocks is trusted and verified, thereby making them a highly secure mechanism for maintaining a distributed ledger of transactions. (p. 2)

Distributed ledger technology (DLT) is the essence of this methodology, although some commentators distinguish between DLT and a true blockchain. Although blockchains were first applied in the financial sector, their potential is being investigated in many other areas, including some aspects of geomatics. The basic format of a blockchain is shown in Fig. 1. Without getting too technical, each block contains a chronological block number, a number used only once or nonce, the data for that block (new addition, requiring verification), and two hash fields (a new hash and a copy of the hash from the previous block) that effectively form the chain. The hash acts as a digital fingerprint in that, if any information within the block is changed, the hash will also change, thus not matching the previous hash in the next block and invalidating the entire blockchain (Kwong, 2019). This is the first stage of tampering proofing, though other mechanisms (such as proof of work ) are also invoked.

 Fig. 1. The basic concept of a blockchain (originally block chain). The nonce is a unique number that ensures the block’s hash meets certain criteria. Note the hash of a block must match the previous hash of the adjacent block.

Blockchain vs Machine Learning

Machine learning is one of the current hot topics in photogrammetry. Indeed, Schneider et al. (2018, p. 341), reporting on the 2018 ISPRS Commission II Midterm Symposium, stated:

“one topic dominated the symposium: machine learning. The incorporation of machine learning as a new tool for increasing the efficiency of the photogrammetric pipeline was evident within many of the presentations and papers”. Much has been made of the difference between: (i) digital platforms that use machine learning (and artificial intelligence AI) such as almost all of the tech giants listed in the first paragraph; and (ii) those that adopt blockchain technology – such as Bitcoin and Ethereum. The argument centers around differences between centralised, decentralised and distributed corporate structures (Fig. 2). Regarding the tech giants Vergne (2020, p. 2) notes: Decentralization may have become a corporate cool factor associated with innovativeness or nimbleness ” yet questions whether the likes of Facebook and Google are truly decentralised:

With their growth accelerating due to an inescapable digitalization trend, boosted even further by pandemic lockdowns around the world, the thought of living in a fully platformized society evokes utopia for some – and dystopia for others. A worst-case scenario would be to have unaccountable corporate behemoths form a platform oligopoly with global surveillance and behavioral prediction capabilities.

FIG . 2. Differences between corporate structures. (a) Centralised structure of a traditional organisation. (b) Decentralised structure claimed by tech giants. (c) Distributed structure of blockchain organisations.

From Vergne (2020).

Companies such as Google (one of Vergne’s corporate behemoths ) rely on machine learning and prides itself on being an ‘AI-first’ company. Machine learning prefers centralized communications to exploit large datasets: as the data builds mass, it attracts further data ( data gravity ). Forstner (in Granshaw et al., 2017) noted that this gravity also extends to personnel, with at least 10 % of the papers at computer vision conferences presented by researchers from the tech giants. On the other hand, organizations that rely on blockchain technology adopt distributed strategies (Fig. 2(c)) and use redundant data as checks, with authentication depending on the collective self-interest of the various participants. Such a methodology has an inherent advantage for open-source applications by providing platform neutrality and consensual decision-making.

Examples of Geospatial Blockchain Applications

It is difficult to find fully photogrammetric applications using blockchain technology at present, yet the potential is apparent in related geospatial disciplines. Dowman (2018) and Ellehauge (2016) list applications including public-good data (for example, street maps), Internet of Things (IoT, for example, drones negotiating air space), land ownership, and geodesic grids. Mendi and C ß abuk (2020) again note land ownership/registration but add border violations and food tracking.

Most land registration applications are designed to circumvent fraud and irregularities, such as in Honduras and Brazil, or to avoid corrupt governments or expensive lawyers in some developing countries. However, Sweden, with one of the most respected land registration systems in the world, has introduced a blockchain land registration pilot to improve transparency and efficiency, rather than to prevent fraud or corruption.

Blockchain in Agricultural Remote Sensing Pincheira et al. (2020) outline potential blockchain uses in remote sensing, where agricultural (arable) applications may adopt multitemporal images from satellites, IoT sensors recording temperature and humidity, plus mobile phone and unmanned aircraft system (UAS) close-range imagery.

However, the owners of the non-satellite imagery (which may be utilized as ground truth) are usually unknown to the other participants, and the data may be of unknown quality (parallels may be drawn with crowdsourcing in applications such as OpenStreetMap).

Currently, this may demand the use of a central intermediary authority which, however, also introduces concerns on ownership, access and integrity and a return to the centralized structure of Fig. 2(a). Using blockchain architecture enables data sharing without an intermediary, tracks updates, provides a quality score and overcomes issues associated with untrusted data owners, using smart contracts to manage relationships in such a distributed structure (Fig. 2(c)).

Blockchain in Navigation

Perhaps the best example of the potential of blockchain technology that may translate to photogrammetry and mapping is by Buttgenbach (2018, p. 18) about marine navigation. He draws a comparison with a ship’s log, but one that is very difficult to falsify: “ A ‘ blockchain ’ is nothing else than a public ledger (a logbook openly accessible to anyone) that exists in multiple copies spread over the Internet, and of which each copy has identical content. Buttgenbach continues: Until now, it was not easy to assess the reliability of a (sea) chart, one had to rely on the word of the hydrographic of offices (the reliability diagrams for example) or of a chartmaker in general. ” Proposing the adoption of blockchain technology for navigation:

The use of DLT when compiling a chart will lay open the originator of the survey data which went into the chart, its time stamp, the identity of the chart compiler, and the means of the generalisation used. . . .Furthermore, it is to be expected that independent, private chartmakers will adopt this technology much faster than the hydrographic of offices. This will create a completely new situation by making commercial charts certifiable and classifiable. . . . Speaking of the Caribbean, modern cruise ships would not be able to enter small ports without tailor-made charts that have not been available from the relevant hydrographic offices and were contracted out to private chartmakers. . . . These ships have to take legal risks using such charts because of the SOLAS (Safety of Life at Sea) regulations that stipulate charts (must be produced only) by hydrographic offices, ignoring the fact that the very same ships would run a risk of running aground using them exclusively. . . .

DLT technology offers a way forward for a renaissance of the profession of private chartmakers working shoulder to shoulder with hydrographic of offices who are free to join the effort on the same basis (p. 19).

Blockchain Implications for Photogrammetry and Mapping

Both the agricultural remote sensing and navigation examples show the potential of blockchain technology for our own discipline. Rather than increasingly turning to the resources of Vergne’s “ unaccountable corporate behemoths ” (who has not used Google Earth?) or, alternatively, the vagaries of crowdsourced data such as OpenStreetMap, we should recognize that our discipline is a heterogenous mix of large players, such as satellite companies and mapping organizations, and individuals or small groups generating significant 3D data from imagery, lidar, and other sensors. Generating a digital elevation model (DEM), a topographic map of a small area or a 3D city model are examples akin to the agricultural and navigation proposals of Pincheira et al. (2020) and Buttgenbach (2018).

For example, a global DEM such as WorldDEM or SRTM may be supplemented by a locally generated model from UAS imagery with a better resolution than the satellite’s data or to fill data gaps. A topographic map from a national mapping agency (NMA) (akin to a navigation chart from a hydrographic office) could be updated from airborne laser scanning (ALS) data from a small commercial enterprise. A 3D city model might utilize mobile mapping data from different players including the local authority and several small companies, supplemented by structure-from-motion (SfM) models from cultural heritage organizations or even amateur enthusiasts.

The issues are the same as with the agricultural and navigation examples: what trust is placed on the efficacy of the data from non-authenticated sources and how will large organizations (such as satellite organizations, NMAs, or the city council) respond? The adoption of blockchain architecture circumvents the intervention of a central intermediary (provided the large players agree). The system and the mass collaboration between its users validate, or reject, the efficacy of the new data that enter the blockchain as a new block. Many players have recently demonstrated an interest: for example, the European Space Agency issued a white paper on Blockchain and Earth Observation in April 2019.

Conclusions

Blockchain technology is very much in its infancy but the widespread adoption of machine learning and AI, as favored by the tech giants, is not the only course for future developments. The trend towards big data and machine learning may point in one direction but the adoption of open-source methodologies and open data in geomatics make the consideration of blockchain technology more than just the domain of cryptocurrencies or land registration. Sensor fusion is associated with disparate data sources from a heterogeneous group of stakeholders with an array of reliability and accuracy issues. Machine learning and AI will continue to be important and dominate developments, with or without the tech giants. But we can also have another string to our bow that embraces the sharing of data from smaller enterprises in a consensual, collaborative, and secure way in the pursuit of open science. Blockchain is not a panacea, but neither should it be considered irrelevant.

STUART I. GRANSHAW
January 2021, Harlech, Wales