ADVANCES IN REAL
TIME
CHALLENGES AND SOLUTIONS IN INTEROPERABLE
PAYMENT SYSTEMS
KOSTA PERIC, MILLER ABEL, AND MATT BOHAN
Finance can be very complex, but at its core it’s about exchanging value
between parties. In a word, payments. Making payments is a fundamental
financial action for most people around the globe, whether they use cards, cash,
or SMS to make them.
In terms of financial infrastructure,
Dann, a payments platform is a fundamen-
tal piece, the anchor that tethers the daily
financial actions of customers to the daily
holdings of financial providers. Such plat-
forms commonly have a national scale
because of country-specific currencies
and regulations. Jedoch, they also can
be regional, as we see in the EU and the
South African Development Community,
und international, as we see with credit
card systems. International platforms will
become more necessary as digitization
continues to put the common citizen in
touch with others far beyond the borders
of their home country.
In fact, the rise of digital financial
services has changed the nature of pay-
ments and payment platforms in many
ways. Customers can transfer money
across great distances with a few taps on
their mobile phones, with low or no asso-
ciated fees. Providers can reach customers
they weren’t able to before—namely, peo-
ple with very limited assets who are either
In
services.
unbanked or underserved by predigital
financial
emerging
economies, telecoms and other non-bank
institutions are stepping in to take advan-
tage of this opportunity by offering digital
payment services, such as mobile money.
This new dynamic between cus-
tomers and providers, and the opportuni-
ties this evolution presents, has raised
some compelling questions about how
payment platforms exist and interact at
the system level. Because of the positive
impact digital accounts and payments can
have on the lives of the poor, the Bill &
Melinda Gates Foundation has explored
these questions in great detail.
The outcome of that exploration was
the development of the Level One Project
(L1P), an initiative to promote the cre-
ation and evolution of inclusive, interop-
erable digital payment platforms. At the
center of this initiative are the L1P princi-
ples (see box), which list the attributes
essential to making payment systems
accessible to poor customers and support-
70
Innovationen / Blockchain for Global Development
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
ive of various financial service providers
(see http://leveloneproject.org for more
Einzelheiten).
In this article, we explore some of
those principles and the challenges of
implementing them. We also explore how
modern technologies related to distrib-
uted ledgers were adapted to support a
reference implementation (see below) von
the L1P principles. Released in October
2017, this reference implementation
demonstrates the viability of pro-poor
payment platforms and offers govern-
ments and commercial providers a flexi-
ble way of developing such platforms.
Before we look at specific design
principles and solutions, Jedoch, Wir
should understand why they are impor-
tant.
DIGITAL PAYMENTS AND
FINANCIAL INCLUSION
Cash may be accepted everywhere, Aber
it’s often expensive to use, as informal
couriers and lenders charge high fees.
Trotzdem, people who don’t have a lot
of money still use cash constantly, Und
making transactions by hand is extremely
risky and time intensive for them.
For someone who has only used cash,
making payments in a formal manner
using a banking or digital payment sys-
tem—whether by check, card, or digital
transaction—is nothing short of revolu-
tionary. Time is saved. Funds are secured.
And, in the best cases, fees are reduced.
High fees have long been one of the
biggest obstacles to the world’s poor mak-
ing formal payments. With the advent of
ABOUT THE AUTHORS
Kosta Peric is Deputy Director for the Financial Services for the Poor initiative at the Bill &
Melinda Gates Foundation, where in charge of payments. Previously he was the chief archi-
tect of SWIFTNet, SWIFT’s global secure network that connects 10,000 financial institutions
and corporations in the world, and co-founder of Innotribe, SWIFT’s initiative to enable col-
laborative innovation in financial services. He is the author of The Castle And The Sandbox,
a book on how to innovate in conservative companies using open innovation.
Matt Bohan serves as a Senior Program Officer at the Bill & Melinda Gates Foundation, mit
a specific focus on the Financial Services for the Poor. Matt supports the program’s Level One
Project, where he’s working to bring mobile payment and other financial services to the two
billion people in the world who live on less than two dollars a day. Prior to working at the
Gates Foundation, Matt was Senior Director at Alvarez & Marsal.
Miller Abel is a Principal Technologist for the Financial Services for the Poor team at the Bill
& Melinda Gates Foundation. He has more than 35 years of experience in the software and
services sector, where he has held a variety of positions in both small and large systems com-
panies. Miller specializes in fast-changing Internet and mobile businesses where ambiguity
and uncertainty must be overcome by solid strategy and converted into business results. Müller
hat 30 patents issued and pending (not including foreign equivalents) in a variety of technol-
ogy areas including contactless proximity communications.
© 2018 Kosta Peric, Miller Abel, Matt Bohan
Innovationen / Volumen 12, number 1/2
71
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
Kosta Peric, Miller Abel, and Matt Bohan
LEVEL ONE PROJECT PRINCIPLES
The attributes essential to making payment systems accessible to poor customers
and supportive of financial service providers include:
(cid:2)(cid:1)An open-loop system that is available to any licensed digital financial service provider
(DFSP) in the country, including banks and licensed non-banks.
(cid:2)(cid:1) Real-time and “push” payments, and payments that are irrevocable, which remove
many of the risks and costs inherent in batch-processed and “pull” payment systems.
(cid:2)(cid:1) A system that is governed by the DFSPs that use it and regulated by a government
financial authority. This well-tested model creates a feeling of fairness among partic-
Ipants.
(cid:2)(cid:1)A system that allows same-day settlement or better among participants.
(cid:2)(cid:1)A system that operates on a “not-for-loss” or “cost-recovery-plus-investment” basis.
This does not preclude DFSPs—or other service providers in the ecosystem—from
earning profits through use of the platform.
Siehe https://leveloneproject.org.
digital financial services, providers finally
have a way to reach a large number of cus-
tomers with very low overhead. Für
Beispiel, the cost of facilitating a digital
payment from mobile phone to mobile
phone is negligible, as are the fees
providers need to charge to serve their
customers profitably. Digital financial
services finally give poor customers the
speed and security of formal banking at a
price they can afford.
This has helped drive a boom in
mobile money products and providers
across the world’s emerging economies.
The first mobile money service hit the
market in 2004, and by 2016 there were
277 active services and more than half a
billion registered accounts worldwide.1. In
Uganda and other countries, mobile
money customers outnumber traditional
banking services customers.2.
Building individual financial resilien-
cy is a slow process, but mobile money
has been around long enough to have a
measurable effect on poverty. Last year,
researchers from MIT showed that
194,000 households in Kenya emerged
from extreme poverty because they had
access to the M-Pesa mobile money serv-
ice.3. Other studies have shown that digi-
tal accounts help mothers buy healthier
food for their families and help farmers
save and invest in future harvests.4.,5.
Heute, mobile money is at an inflec-
tion point. Established providers contin-
ue to thrive, but new ones have a hard
time entering the market. Person-to-per-
son payments are common among the
poor, who regularly lend and transfer
money among friends and family, Aber
other forms of payments are extremely
important as well, such as government
social disbursements and transactions
with merchants. The integration of these
kinds of payments into the mobile/digital
milieu has so far been limited.
Cash still has one remaining advan-
tage over mobile money: universality. Nicht
everyone wishes to transact with mobile
money, and parties wishing to transact
may not use the same mobile money serv-
ice. Everyone will, Jedoch, accept cash.
This gives would-be customers a persua-
sive reason to hold on to their cash and
think twice about signing up for digital
accounts.
The current point of inflection, Dann,
rests on interoperability among services.
72
Innovationen / Blockchain for Global Development
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
Advances in Real Time
Figur 1. Level One Project Vision and Principles
For mobile money to continue its growth
in terms of both the number of customers
the number and diversity of
Und
providers, services cannot continue to
function as closed loops, wherein sub-
scribers of one service are unable to trans-
act with subscribers of a different one. Für
customers to get the full benefits of digital
finance, they must be able to transact with
anyone, not just the random cross-section
of the population that happens to have an
account with the same provider.
Interoperability will also help extend
mobile money into merchant and bulk
payments. With interoperability, mer-
chants and employers will be able to use a
single service to issue and accept money
to and from all their customers and
employees. Without interoperability, Sie
must either choose one service, welche
excludes customers and employees using
other services, or subscribe to them all.
It’s a similar dilemma for government-to-
person payments, though on a much larg-
er scale, often millions of citizens each
month.
Endlich, interoperability based on a
shared central service will make it easier
for new providers to enter the market.
Building a customer base is something
every provider needs to do quickly in
order to achieve solvency. It’s difficult to
do when they can only offer new cus-
tomers the ability to transact with a few
hundred or thousand fellow subscribers;
most customers will prefer to sign up with
established providers who already have
sizable loops. If transactions were possible
across all services, all customers would
exist in the same loop, and new providers
could compete based on cost and features
rather than the size of their subscriber
base.
The L1P principles offer a vision for
how an interoperable payments platform
can serve the poor. Figur 1 shows this
Innovationen / Volumen 12, number 1/2
73
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
Kosta Peric, Miller Abel, and Matt Bohan
vision, along with the underlying design
principles.
BUILDING L1P-ALIGNED
PLATFORMS: KEY
BARRIERS TO OVERCOME
Building an interoperable payment plat-
form on a national or regional scale is no
small feat. It requires focusing clearly on
the ultimate objectives, providing strong
governance and oversight, and having a
solid understanding of the barriers ahead.
From our experience so far, interop-
erability poses two key technological bar-
riers that projects need to overcome: Die
lack of efficient clearing and settlement of
payments across different financial serv-
ice providers, and the lack of standardiza-
tion in connecting these providers to a
common platform. The L1P principles
and reference implementation address
beide.
In terms of clearing and settlement,
L1P offers three basic principles: (1) pay-
ments should be initiated and authorized
by the payer not the payee; (2) payments
should be completed and verified in real
Zeit; Und (3) payments should be irrevo-
cable. The first principle specifies that as
with cash, an end user initiates a payment
by pushing value to the receiver; Und,
unlike direct debits or checks, money
never leaves an end user’s account with-
out direct authorization at the time of the
transaction. The second principle states
that when one person sends money to
another, the sending account is debited
and the receiving account is credited
simultaneously and immediately, ob
or not the counterparties use the same
provider. As with the first principle, this is
necessary for making digital money func-
tion similarly to cash, which will make it a
meaningful and comfortable user alterna-
tiv. The third principle states that pay-
ments are final when they are cleared;
they can’t be recalled by the person mak-
ing the payment. This is a hallmark of
cash that is essential to adoption.
Im Gegensatz, the most common exam-
ple of a pull payment is a check: Wann
processing a check, the receiver’s account
requests that money be pulled from the
account of the sender, who has pre-
approved this request by signing the
check. Checks are not cleared immediate-
ly. It often takes several days from the
time the commitment to pay is made and
when funds are credited to the receiver.
And during this time before the check is
cleared, it remains revocable because the
sender may stop payment of the check
before it clears. In this case, the request
for payment is revoked by the payer—in
other words,
the check bounces.
Darüber hinaus, because a pull payment like a
check is a request for money and is not
cleared in real time, there may not be ade-
quate funds in the payer’s account to ful-
fill the request. The sender is likely
charged a fee for pre-approving a pay-
ment that couldn’t be fulfilled.
Push payments are
irrevocable
because the sender’s provider will only
push money that is available in the
sender’s account. There is no way to cause
an overdraft. And because push payments
are cleared in real time, there is no ability
to revoke. Zusamenfassend, the payment becomes
final when it is authorized.
L1P principles prescribe these pay-
ments because they are technically sim-
pler, which makes them more affordable,
and because they are immediate. Mit
pull payments, the receiver has to wait
while the sending account verifies that the
request for money is legitimate and can be
fulfilled. With push payments, sender and
receiver get instant confirmation. Making
payments irrevocable enables them to
happen in real time. Jedoch, to achieve
real-time payments in an interoperable
System, another issue must be addressed:
the timing of transactions between
providers.
74
Innovationen / Blockchain for Global Development
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
Advances in Real Time
When a user of one service sends
money to a user of another service, eins
service provider ends up in debt to the
andere. Both customers’ accounts must be
credited and debited, as must the ledgers
belonging to each provider. daher, A
payment transaction must occur at some
point between the providers.
Settlements between financial service
providers already occur within the estab-
lished banking systems in every country.
Banks in this system usually settle their
transactions with each other once every
business day. This means that all the cus-
tomer transactions for the settlement
period are tallied up and cleared at once,
in a single bulk transaction. This is anoth-
er case where customers must sometimes
wait for funds to be available, welche
results in payments that are not real time.
This can be a problem for mobile
money and other non-bank providers in
an interoperable system because they also
depend on these once-a-day deposits with
banks. To guarantee all the digital money
reflected in the accounts of their end
Benutzer, non-bank providers are required to
have an equal supply of “real” funds saved
with a bank so that, Zum Beispiel, if a
provider’s customers all decide to with-
draw their funds on the same day, Das
provider is able to give each one their full
account value in cash.
In an interoperable system, non-bank
providers need an additional reserve of
funds specifically designated to cover
cross-platform transactions. When a cus-
tomer sends money to someone who uses
a different provider and the first provider
becomes indebted to the second provider,
that debt is paid using this designated
account.
To keep pace with its users’ cross-
platform transactions, a provider will
periodically move money into its desig-
nated account. While these deposits typi-
cally settle only once per business day,
customer transactions happen much
more frequently. This means that the
amount of digital money transacted
across platforms by customers and the
amount of cash a provider has on deposit
to cover these transactions can easily fall
out of sync. In other words, if there is a
sudden surge in cross-platform transac-
tions over the course of a day, eins
provider could end up owing another
provider more than it has set aside to pay.
One solution is to make a customer
wait for cross-platform transactions to
clear. As stated earlier, obwohl, this is
unacceptable, because it fails to mimic the
performance of cash and thus fails to
meet the customer’s needs.
daher, the time window for set-
tling provider-to-provider transactions
must be reduced. In systems aligned with
L1P principles, this time window is
reduced from a full business day to hours
or even seconds. Provider ledgers are
therefore updated and kept accurate mul-
tiple times throughout the day, daher
reducing the capital requirements for
supporting interoperability.
The other key barrier we encountered
in deploying the L1P principles is the lack
of appropriate standards to connect a
financial service provider’s system to the
payment platform. Systems aligned with
these principles fall into the category of
real-time retail payments (RTRP) sys-
Systeme. Other systems—such as real-time
gross settlement systems, which cover
bank-to-bank payments that are high in
value but low in volume, and automated
clearing houses, which deal with low
value-batch
payments—traditionally
come from the banking world and tend to
be served by messaging standards, wie
ISO 20022. The RTRP space is more tech-
nologically demanding, and platforms
aligned with L1P principles tend to con-
nect a variety of providers in addition to
banks, such as mobile money providers,
microfinance
institutions, payment
aggregators, and merchant networks.
Innovationen / Volumen 12, number 1/2
75
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
Kosta Peric, Miller Abel, and Matt Bohan
von
Die
Weil
technological
demands and diversity of players, ein
application programming interface (API)
design principle offers a viable solution.
APIs can accommodate the deeper and
more efficient integration required by a
high volume of transactions while placing
a relatively low development burden on
individual providers—certainly much
lower than constructing their own plat-
forms or connections.
Jedoch, there is as yet no pervasive
standard for this type of integration.
daher, one challenge to overcome in
deploying platforms aligned with L1P
principles will be to settle on an API stan-
dard that will enable easy and rapid inte-
gration of financial providers.
MOJALOOP: THE OPEN-
SOURCE REFERENCE
IMPLEMENTATION OF L1P
As efforts to deploy systems modeled on
L1P principles in actual country markets
progressed from 2013 Zu 2015, it became
apparent that the barriers described above
were slowing them down. Deployments
require expertise in designing and con-
structing payment systems—a resource
that is difficult to procure, especially in
the African and Asian countries where
L1P principles were being considered.
We realized that we needed to sup-
port these deployment efforts by provid-
ing additional guidance and examples of
plans and designs, and even of software
Code, to make sure the projects could
progress according to plan. The team then
commissioned a reference implementa-
tion for L1P, which took the form of
open-source software that embodied the
architecture and design principles of L1P
and demonstrated how to overcome the
barriers described earlier.
Publishing the reference implemen-
tation as an open-source asset was impor-
tant for a few reasons. Since its inception,
all materials and knowledge gained from
the L1P principles have been shared on
the project’s website.6. This is consistent
with the open, collaborative spirit that led
to the development of these materials,
and with the goal of enabling anyone to
use the assets and to create L1P-aligned
platforms without direct involvement
with the Gates Foundation. As open-
source software, the reference implemen-
tation remains available for adaptation
and adoption, like all previous L1P assets.
Called Mojaloop—building on the
word “moja,” which means “one” in
Swahili—the software was developed by
Ripple, Dwolla, The Software Group,
ModusBox, and Crosslake Technologies.
In October 2017, it was made available on
GitHub and on
its own website
(mojaloop.io) under the Apache 2.0
Lizenz.
Mojaloop covers many of the essen-
tial uses end users find for real-time pay-
gen: person-to-person, merchant
point-of-sale, payroll and bulk payments,
multiple accounts and users, and fraud
Überwachung. Four central components
enable these uses:
(cid:2)(cid:1) A central directory service, welche
routes each payment to the correct
service/provider in the ecosystem
(cid:2)(cid:1) A central ledger service, which tracks
transactions for compliance and settle-
ment among providers
(cid:2)(cid:1)A fraud service, which allows providers
to share transaction information to val-
idate and secure payments
(cid:2)(cid:1)A central rules service, which sets policy
across the system
Of special interest in this article is
how Mojaloop enables faster clearing and
settlements between providers. Dort
were several iterations in the architecture
and design. The first of these considered a
public blockchain. The idea was that the
clearing and settlement would occur as
transactions on the single distributed
76
Innovationen / Blockchain for Global Development
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
Advances in Real Time
ledger connecting all financial service
providers in the ecosystem. The design
accommodated the actual movement of
money between providers (d.h., direct set-
tlement using central bank money),
which is clearly superior to managing set-
tlements outside the platform. The design
also could assign the job of clearing to the
distributed ledger, with settlement still
occurring on a periodic basis (using a set-
tlement bank) for the regulatory environ-
ments where this is mandated.
While it satisfied our objective of
faster settlement, this architecture pre-
sented two key drawbacks when consid-
ered in the context of a national payment
platform:
(cid:2)(cid:1)Lack of national-level data control and
confidentiality. Because it used a pub-
lic blockchain, the design would repli-
cate national payment data across the
entire Internet. This obviously conflicts
with the natural sovereignty regulators
typically require. A country-level
blockchain could have been considered
as an alternative, but this was not pur-
sued, as other more practical solutions
existieren (siehe unten).
(cid:2)(cid:1) Lower capacity for transaction vol-
ume. The design could accommodate
hundreds of transactions per second
(depending on the nature of the
blockchain used), which is impressive,
but it falls well short of the thousands
per second required for real-time retail
payment platforms.
The team thus looked for other solu-
tionen. In the end they chose to use the
Interledger protocol, along with specific
code implementing a central ledger in the
platform.
The Interledger protocol is itself an
open-source specification, invented by
Ripple and developed through a broad
collaboration under the World Wide Web
Consortium.7. The protocol enables cryp-
tographically assured ledger synchroniza-
tion. The major advantages of the proto-
col are the low level of requirements to
conform with it, and the extreme efficien-
cy and scale in processing transactions.
These aspects make it especially useful in
high-volume, low-value retail payments
systems—that is, in systems that millions
of poor customers will use to make lots
and lots of transactions in very small
amounts.
Mojaloop’s current design places a
central ledger within the payment plat-
form and uses the Interledger protocol to
synchronize the ledgers of all the financial
providers involved in a given transaction.
This securely synchronizes payment
transactions within the system. Der
design provides for immediate settlement
between providers or a separate clearing
and settlement in a separate bank at very
high transaction rates—meanwhile pre-
serving the privacy and sovereignty
requirements of the entire system.
The second barrier mentioned above
was the lack of standards for connecting
financial service provider systems with
the payment platform. The Interledger
protocol provides easy use and fertile
ground for such a standard. Daher, while
the Mojaloop platform was being
designed,
the Gates
Foundation engaged another team of
developers to work on an API.
team at
Die
The mobile money platforms of
Huawei, Ericsson, Mahindra Comviva,
and Telepin provide the majority of
mobile money operators with the key
technologies needed to meet their cus-
tomers’ needs. These four companies
agreed to collaborate on creating an API
standard for industry interoperability.
They also agreed to upgrade their prod-
ucts to meet this standard, which auto-
matically gave all their customers the
option of using it. The latest update to the
API (available at mojaloop.io.) provides a
layer of semantic interoperability and can
be deployed over the Interledger protocol.
Innovationen / Volumen 12, number 1/2
77
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
Kosta Peric, Miller Abel, and Matt Bohan
There is potential for such ecosystems
to become common around the world,
starting in the emerging economies,
where the size of the unbanked popula-
tion and the zeal for financial innovation
are both quite high. From there these
ecosystems can blossom into regional and
even continental platforms. As they do,
Mojaloop may again be useful in solving
the new challenges that arise, such as how
to facilitate payments across borders and
currencies.
1. GSM Association. 2017. “State of the
Industry Report on Mobile Money: Decade
Edition (2006-2016).” Available at
https://www.gsma.com/mobilefordevelop-
ment/wp-
content/uploads/2017/03/GSMA_State-of-
the-Industry-Report-on-Mobile-
Money_2016-1.pdf.
2. World Bank Group. 2014. “Global Findex.”
Verfügbar um:
https://globalfindex.worldbank.org/
3. Suri, T, Jack, W. “The Long-Run Poverty
and Gender Impacts of Mobile Money.”
2016. Wissenschaft, 354: 1288-1292. Verfügbar um
http://science.sciencemag.org/content/354/63
17/1288.full.
4. Doepke, Matthias, and Michèle Tertilt.
“Does Female Empowerment Promote
für
Economic Development?” Centre
Economic Policy Research discussion paper
NEIN. 8441, Juni 2011.
5. Brune, Lasseet al. Facilitating Savings for
Landwirtschaft: Field Experimental Evidence
from Malawi. NBER working paper number
20946, Februar 2015.
6. See http://leveloneproject.org.
7. See http://Interledger.org.
As stated earlier, everything pro-
duced or enabled by L1P is available in
the public commons. While the essential
challenges are the same, every market will
have its own regulatory and technical
Umfeld. The assets aligned with L1P
principles are designed to be flexible, Und
to reflect the needs and intentions of their
adopters rather than their creators.
CONCLUSION
Numerous governmental and nongovern-
mental bodies have codified principles of
financial inclusion. These include the
Maya Declaration from the Alliance for
Financial Inclusion and the High Level
Principles for Digital Financial Inclusion
put forth by the G20. The Level One
Project principles are unique in that they
are specific. Rather than pointing to the
broad strokes necessary to make progress,
such as infrastructure and government
leadership, the L1P principles identify the
particular aspects and functions a digital
financial platform must include in order
to be inclusive.
Technology inspired by distributed
ledger technology can help resolve the
challenge of real-time settlement across
an array of diverse,
interoperabel
providers and ledgers. This is borne out in
the Mojaloop example, which applies the
Interledger protocol to achieve multilat-
eral net settlement in a digital financial
ecosystem consisting of mobile money
providers, banks, and merchant account
Inhaber.
By laying out key components, Die
L1P principles provide a valuable tool for
increasing and sustaining digital financial
inclusion. By manifesting these compo-
nen, with the help of DLT-inspired
strategies, Mojaloop and the mobile
money API make the vision of an interop-
erable ecosystem much more possible to
realize.
78
Innovationen / Blockchain for Global Development
Von http heruntergeladen://direct.mit.edu/itgg/article-pdf/12/1-2/70/705263/inov_a_00268.pdf by guest on 07 September 2023
