Search, Find and Resolve: Towards a Taxonomy for Searchable

Encryption Schemes

Ines Kramer, Silvia Schmidt, Mathias Tausig and Manuel Koschuch

Competence Centre for IT Security, University of Applied Sciences FH Campus Wien, Vienna, Austria

Keywords:

Searchable Encryption, Taxonomy, Inverted Index, Forward Index, Full-domain Search.

Abstract:

Searchable Encryption (SE) schemes are a promising solution to the problem of outsourcing one’s data to

a cloud provider in a secure way, while still retaining the ability to search for and easily retrieve speciﬁc

documents. A multitude of different schemes have been proposed and designed, yet in general they still lack

usability/applicability for a speciﬁc use case or proper security analysis in order to be widely implemented and

used. To address this issue we started a project to determine which SE schemes ﬁt certain use cases - mainly

focusing on usability. We examined nearly 400 papers on SE schemes from the last 13 years and extracted

categorization domains for SE schemes. Furthermore we took a time-based look at these domains and tried

to identify future trends in SE technologies. In this position paper we introduce our methodology and give a

short overview of our current work-in-progress.

1 INTRODUCTION AND

RELATED WORK

Since Song et al. published their seminal work in

2000 (Song et al., 2000), research on Searchable En-

cryption (SE) delivered numerous articles and diverse

SE schemes. Multiple improvements regarding us-

ability and security were achieved, while also new

threat models and attack scenarios were discovered.

In addition to that, papers with various approaches

towards establishing an SE taxonomy or respective

surveys have been published. The ﬁrst survey-like

article was released by Sedghi et al. (Sedghi et al.,

2008) in 2008 and proposed a categorization based on

information leakage, i.e. information leakage while

writing data vs. information leakage while perform-

ing search on data.

In 2011 Boneh et al. (Boneh et al., 2011) provided

deﬁnitions on Functional Encryption, which were di-

vided into Predicate Encryption and Predicate En-

cryption with Public Key. The taxonomy by Tang

(Tang, 2012) was published the following year. He

deﬁned four main SE categories based on the kind of

encryption and what kind of search is performed i.e.

asymmetric and symmetric encryption - each divided

into index-based and full-domain search.

A categorization based on functionalities of

https://orcid.org/0000-0001-8090-3784

Privacy-Assured SE schemes was published by Li

et al. in 2013 (M. Li et al., 2013). They used a

top-down methodology for designing privacy-assured

search schemes: search functionality → information

retrieval → data index structure → primitive data op-

eration → cryptographic design → SE primitives.

The most comprehensive survey on SE to date

was written by B

osch et al. (B

osch et al., 2014).

They mainly differentiate between asymmetric and

symmetric encryption. A further categorization by

means of single or multi writer(s)/user(s) was intro-

duced, which led to four classes in the form of SE

Writer/Reader: SE Single Writer (S)/Multi Reader

(M), SE S/S, SE M/M, and SE M/S.

The more recent position paper by Cui et.al. (Cui

et al., 2017) analyses 24 schemes according to their

leakage patterns and security requirements. Any SE

scheme suffers from a form of information leakage

to the server, which is described by leakage patterns.

The index/size pattern refers to the information that

can be deduced from the stored ciphertext/index. This

includes documents or index size, number of docu-

ments or keywords, document lengths and similarity.

The search pattern reveals the same keyword by com-

paring the matched records of two queries. By exam-

ining the history of result-sets in repeated queries the

access pattern can be inferred.

Based on these articles we categorized recent

SE work by tagging almost 400 papers mainly re-

414

Kramer, I., Schmidt, S., Tausig, M. and Koschuch, M.

Search, Find and Resolve: Towards a Taxonomy for Searchable Encryption Schemes.

DOI: 10.5220/0007753804140421

In Proceedings of the 4th International Conference on Internet of Things, Big Data and Security (IoTBDS 2019), pages 414-421

ISBN: 978-989-758-369-8

trieved from IEEE, IACR, and other (preferably peer-

reviewed) online repositories. Our initial motivation

was to ﬁnd a scheme suitable for implementation in

an existing cloud system. Soon we realized that we

had to start by categorizing the existing schemes and

by doing so were able to identify various directions of

SE research.

In general, SE research focuses on three aspects

osch et al., 2014):

• Efﬁciency

• Security

• Query expressiveness

Our main focus is set on identifying and imple-

menting a usable and well-performing (by means of

runtime and memory) scheme for practical use. With

this in mind, Tang’s (Tang, 2012) taxonomy lacks

categories for practical applications, e.g. number of

users with read and write access privileges.

Therefore the approach by B

osch et al. (B

osch

et al., 2014) is convincing in terms of determining an

SE scheme for practical usage.

SE has to focus on three different trade-off scenar-

ios (B

osch et al., 2014):

• security vs. efﬁciency

• query expressiveness vs. efﬁciency

• security vs. query expressiveness

When increasing the security of a scheme, efﬁ-

ciency almost always suffers. The less secure the sys-

tem is built, the more efﬁcient it usually can perform.

Furthermore, SE schemes should offer high expres-

siveness in search options, which often leads to less

secure and/or less efﬁcient systems. It depends on

the speciﬁc individual application scenario which key

ﬁgures have to be prioritized: Can the content of the

dataset be derived by domain knowledge, and which

level of security is required? How big is the underly-

ing dataset, and is there a need for dynamic changes?

How often will queries be performed? Thus, elabo-

rating certain criteria regarding SE research (mainly

attributes of SE schemes) is indispensable.

In this position paper we want to give some in-

sights into our current work-in-progress. Since we

try to cover the majority of work published since

SE schemes were initially introduced, this paper cur-

rently only discusses our intermediate results up to

papers published during 2016. We are aware that this

discussion is currently severely lacking works pub-

lished in the last 2 years (IEEE alone lists almost 170

publications for this period, as much as in 2014 to

2016 combined), but we are currently actively work-

ing on ﬁnishing our analysis.

1.1 Technical Corpus

We began by dividing the overall 393 articles on SE

into 6 topics, see Table 1.

Table 1: Segmentation of the surveyed papers into general

topics. Some papers have been assigned to several topics,

so the Ratio column adds up to more than 100%.

Topic Ratio

SE Background Information 23.4%

SE Surveys 2.3%

SE Security Analysis & Attacks 4.6%

SE Applications 4.3%

SE Schemes 66.2%

SE Implementations 31.0%

SE Applications represents work on explicit ap-

plied usage of SE schemes, whereas SE Implementa-

tions contains articles which also provide an evalua-

tion of the implemented SE scheme.

Often, implementations focus on testing efﬁciency

(for a variety of metrics) and security for certain SE

algorithms of the scheme. Therefore, the majority

of implementations do not consider integration into

a client/server framework.

Furthermore, the SE schemes were categorized by

cryptographic settings ((a)symmetric/hybrid), search

settings (full-domain/index-based/hybrid), query ex-

pressiveness, and data types. Several articles focused

on more than one topic.

91 (23.2%) articles of all 393 collected papers focus

on diverse background information for SE, i.e. certain

aspects of SE schemes are elaborated in these articles.

These aspects are given in Table 2 and give a certain

indication of the research trends in underlying primi-

tives.

Table 2: Technical aspects dealt with in the identiﬁed SE

Background Papers.

Technical Aspects Ratio

Bilinear Maps 22.0%

Homomorphic Encryption 18.7%

Identity-Based Encryption 13.2%

Private Information Retrieval 9.9%

Attribute-Based Encryption 7.7%

Lattice-Based Systems 4.4%

Functional Encryption 2.2%

Hidden-Vector Encryption 2.2%

Broadcast Encryption 2.2%

The remainder of this position paper is now struc-

tured as follows: We give an overview on the different

Search, Find and Resolve: Towards a Taxonomy for Searchable Encryption Schemes

415

schemes and their corresponding domains we encoun-

tered during our research in Section 2, followed by the

extraction of an SE-development-timeline in Section

3 in order to ﬁnd out if we can identify certain trends

in the development of SE. We ﬁnally conclude in Sec-

tion 4 by giving a short outlook on our planned future

work.

2 MAIN CATEGORIZATION

DOMAINS

Throughout our research of 393 articles on SE (pub-

lished between 2000 and December 2016) we identi-

ﬁed 260 different SE schemes and the main domains

for categorizing these schemes involved:

• number of users (writer/reader)

• cryptographic primitives

• types of search

• search criteria

• types of data

• security

In the remainder of this Section we will give a

short overview of our ﬁndings for each domain.

2.1 Number of Users (Read/Write

Access Privilege)

Categorizing by means of writer(s) and reader(s)

osch et al., 2014) in an SE scheme is of signiﬁ-

cant importance regarding the practical applicability

of such a scheme. Choosing an SE scheme for a given

purpose immediately raises questions on access priv-

ileges. Who is allowed to write (i.e. generate and

transmit encrypted data to the server) and who is al-

lowed to read (i.e. submit a query by issuing a trap-

door and interpret its results).

In 39 (15%) of 260 papers on SE schemes no

information on the number of users, in terms of

writer(s) and reader(s) was found. The remaining pa-

pers contain

• 72 S/S (27.7%; single writer/single reader) sys-

tems,

• 52 S/M (20%; single writer/multiple readers) sys-

tems,

• 72 M/S (27.7%; multiple writers/single reader)

systems, and

• 34 M/M (13.1%; multiple writers/multiple read-

ers) systems.

All SE schemes with any kind of multiple users

(*/M or M/*) are suitable for sharing data. S/S

schemes are mainly built for outsourcing data for a

single entity. All */M SE settings require some mech-

anism for key distribution and access control to allow

multiuser reading.

2.2 Cryptographic Primitives

When classifying SE schemes regarding their crypto-

graphic primitives, one always looks at the way how

these primitives are used in SE related operations (e.g.

search queries, index) and not how the actual data is

encrypted. So it still possible (and common) for an

asymmetric scheme to work with symmetrically en-

crypted data.

In general, in a symmetric scheme there exists one

secret key. Whoever holds that key can add data to the

repository and perform searches on that data.

In an asymmetric setting, the entities in possession

of the private key can issue search queries and retrieve

data, while possession of the public key allows for

data to be added to the repository.

Currently there doesn’t seem to be a general con-

sensus of when to call a system hybrid, so when giv-

ing this classiﬁcation we followed the one given by

the authors in their original publication.

Among the 260 articles on SE schemes we identi-

ﬁed

• 117 symmetric schemes (45%),

• 132 asymmetric schemes (50.8%), and

• 19 hybrid schemes (7.3%)

(some papers focused on more than one scheme).

The ﬁrst symmetric SE scheme (SSE) published

in 2000 (Song et al., 2000) represents the S/S set-

ting as described in Subsection 2.1. Asymmetric SE

schemes where initially called public key encryption

(PKE) schemes. In 2004 PEKS (Public Key Encryp-

tion with Keyword Search) was introduced by Boneh

et al. (Boneh et al., 2004), originating the current term

PEKS for asymmetric settings and M/S SE schemes.

2.3 Types of Search

There are basically two types of search in SE: the

full-domain search and the index-based search. Both

searches blind the search queries to avoid data leakage

(Tang, 2012).

Full-domain search means that the search is per-

formed over all data, while the search in index-based

systems is performed on pre-deﬁned keywords gath-

ered in an index. The keywords have to be care-

fully selected to allow for meaningful queries, but this

IoTBDS 2019 - 4th International Conference on Internet of Things, Big Data and Security

416

technique is usually less complex and more ﬂexible

in terms of data encryption. Its drawbacks are the

decreased ﬂexibility when updating the index, addi-

tional memory requirements for storing the index, and

the expensive (re-)building of the index.

The index-based search is divided into three sub-

categories: forward index, inverted index, and hybrid

index. In a forward index, each data item is associ-

ated with a list of keywords (cf. Table 3). Worst case

search complexity is quadratic (for n documents with

n keywords each). Yet updating the index (be it by

changing keywords or documents) is usually fast and

easy.

Table 3: Example of a Forward Index (Tang, 2012).

Data Item Index

Document 1 black, red, white, chrome, vienna

Document 2 blue, white, copper, gold, memphis

Document 3 yellow, orange, gold, rome, vienna

An inverted index is one single index represent-

ing all data (cf. Table 4). It is a further possibility

for minimizing search time since it is now linear in

the number of keywords. However updating this in-

dex is usually a more costly and complex operation.

Schemes that allow for efﬁcient updates of the index

are called dynamic, while static schemes do not sup-

port insert, update and delete operations.

Table 4: Example of an Inverted Index (Tang, 2012).

Data Item Index

black Document 1

red Document 1

chrome Document 1

vienna Document 1, Document 3

blue Document 2

white Document 1, Document 2

copper Document 2

memphis Document 2

yellow Document 3

orange Document 3

gold Document 2, Document 3

rome Document 3

The 260 articles on SE schemes contain

• 19 full-domain search systems (12.3%; 12 asym-

metric, 7 symmetric) and

• 129 index-based search systems (87.7%; 49 asym-

metric, 80 symmetric; with 8 hybrid, 13 for-

ward index, 28 inverted index, and the remaining

schemes without an explicitly stated index struc-

ture).

Further we identiﬁed 26 schemes explicitly consid-

ered to be dynamic, 20 of them index based. The ﬁrst

dynamic inverted-index approach has been proposed

in (Liesdonk et al., 2010).

The remainder of the schemes were not explicitly

associated with a certain type of search.

2.4 Search Criteria

Search criteria must not be neglected when evaluating

the usability of an SE scheme. Different applications

have different requirements in terms of query expres-

siveness. Early SE schemes offered search setting for

queries with a single keyword; today most systems

support at the very least multiple keywords.

This categorization also includes the results of a

query, e.g. ranked results. In our review, we identiﬁed

the search criteria given in Table 5.

We additionally encountered the following search

criteria in less than ﬁve schemes:

comparison, semantic search, subset queries, struc-

tured query (SQL), phrase search, and veriﬁable

search.

These low numbers may be caused by the fact that

these kinds of query where only introduced near the

end of our investigated body of work, e.g. structured

query was ﬁrst published in 2016.

By default, SE search queries are equality queries,

thus equality search was not explicitly listed in Table

Boolean queries support boolean operations such

as conjunction, disjunction, and negation on key-

words. Ranked search delivers the query results

ranked by a pre-deﬁned order, for example by rel-

evance to the given keyword (query) (Wang et al.,

2010). In systems with range queries (incl. subset

queries) the search is performed in a certain range

within the data, e.g. the ﬁrst 100 documents. Most

users nowadays are used to fuzzy keyword search

(e.g. Google), where even typos and incomplete sub-

mitted keywords produce the desired search results.

Therefore, fuzzy keyword search is seen as a further

step towards usability - like similarity search.

2.5 Types of Data

A further criterion for categorizing SE schemes is the

kind of data SE is performed on. By default all the SE

schemes are elaborated on simple ﬁle servers contain-

ing text documents, but in our investigated set there

are also

• 26 databases,

Search, Find and Resolve: Towards a Taxonomy for Searchable Encryption Schemes

417

Table 5: Query Types supported in the investigated papers.

Query Type Ratio

Boolean Queries 21.9%

Ranked Results 14.2%

Range Queries 9.2%

Fuzzy Keyword Search 5.4%

Similarity Search 3.1%

Delegated Search 2.3%

• 7 audit logs,

• 5 mobile devices,

• 4 public health records, and

• 2 email settings.

Furthermore, we found geodata, face recognition

systems, genomic data, and images.

2.6 Security

To gain a deeper insight into the security constraints

of SE schemes we evaluated several leakage abuse at-

tacks. These are divided into two main branches ac-

cording to their impact.

• Partial plaintext recovery attacks against schemes

with fully-revealed occurrence pattern with key-

word order (L3) and query-revealed occurrence

pattern (L2) as described in (Cash et al., 2015;

Naveed et al., 2015).

• Query recovery attacks, which reveal the queried

keyword(s). These generally do not depend on the

underlying SE scheme. They also work for query

revealed occurrence pattern schemes (L1), as used

in most of the examined articles.

The ﬁrst Query recovery attack proposed by (Is-

lam et al., 2012) exploits access pattern leakage. With

prior knowledge of the used dataset, the contents of

the search queries can be guessed with high accuracy.

These attacks are even possible if a scheme is proven

to be secure under the standard assumption, which

means these attacks are permitted by default.

Cash et al. increased the efﬁciency of this attack

in (Cash et al., 2015). Their count attack is based on

the fact that a large fraction of keywords will match

against a unique number of documents. Hence, an

adversary who knows the plaintext documents sim-

ply counts the number of documents returned by each

keyword and compares it to the number of documents

matched by a query.

An attack based on the search pattern has been

presented by Liu et al. (Liu et al., 2013). An adver-

sary who has knowledge of the user’s search habit can

effectively attack the keywords underlying the query

with the help of some publicly available knowledge.

Zhang et al. extended the active known document

attacks from (Cash et al., 2015) which rely on the

access pattern and named them ﬁle injection attacks.

The authors propose a non-adaptive and an adaptive

version depending on inserting ﬁles before or after

search queries are made. These powerful attacks work

for all dynamic schemes which are not forward se-

cure, even if they provide a low leakage. An adver-

sary controlling the server is able to learn a very high

fraction of keywords searched by a client using a rel-

atively small number of injected ﬁles. Table 6 gives

an overview of the different leakage attacks currently

deﬁned in the literature.

Table 6: Summary of the different leakage abuse attacks.

(Islam

et al.,

2012)

(Cash

et al.,

2015)

(Liu

et al.,

2013)

(Zhang

et al.,

2016)

Property n.a. n.a. n.a. dynamic

Leakage

Pattern

Access Access Search Access

Required

Infor-

mation

Known

plain-

text

dataset

Partial

known

plain-

text

dataset

Publicly

known

pattern

Injected

ﬁles

Due to the power of ﬁle injection attacks we ex-

tended our classiﬁcation by Forward Privacy and

Backward Privacy.

• Forward Privacy (forward secure): Ensures that

newly added data remains hidden to the server un-

til it gets revealed by a later query, even if the

server might have learned some secrets during

previous queries.

• Backward Privacy (backward secure): Search

queries should not leak matching entries after they

have been deleted.

Just 4 schemes of the 26 explicitly considered dy-

namic ones in the evaluated period of time are consid-

ered forward secure (Chang and Mitzenmacher, 2005;

Stefanov et al., 2013; Bost, 2016; Bost et al., 2016)

and up to 2016 there are none providing backward se-

curity (Cui et al., 2017).

3 DEVELOPMENT TRENDS IN

SEARCHABLE ENCRYPTION

Finally, we connected our results to a timeline to ﬁnd

out if one can identify certain trends in the develop-

IoTBDS 2019 - 4th International Conference on Internet of Things, Big Data and Security

418

ment of SE schemes. The outcome of these time-

based results are described in this section.

We found out that the number of works on S/S and

S/M schemes is increasing during the last 5 years of

our observed set. On the other hand, the all-over trend

for M/S is decreasing; M/M SE remains more or less

on the same level.

Regarding cryptographic primitives we found a

remarkable gap between asymmetric and symmetric

settings from 2005 to 2010. After 2010 both keep on

the same level with a slightly increasing trend. The

ﬁrst works on hybrid schemes were published in 2011

and the number of research work is slightly increas-

ing - Figure 1 illustrates the trends of cryptographic

primitives in SE schemes.

Figure 1: SE based on cryptographic primitives 2003-2016.

The SE search setting itself is signiﬁcantly dom-

inated by index-based settings. Full-domain search

keeps constantly in low numbers. Figure 2 shows

further increase in the future on SE schemes with

an index-based search. The focus of the research is

clearly on the efﬁciency of search queries, deducible

from the increasing trend of using inverted indexes.

Whereas the number of schemes utilizing forward in-

dex settings are slightly decreasing.

Figure 2: Full-domain vs. index-based SE settings 2003-

2016.

We identiﬁed various search techniques, with the

most popular being:

• boolean search,

• range search,

• ranked search, and

• fuzzy search.

The earliest scheme supporting boolean search

queries appeared in 2003, this has been the most

frequently engineered technique until 2014. Then

ranked search became the most popular query expres-

siveness, which provides search results in a certain

order. Ranked and fuzzy keyword search initially

appeared in our paper collection in 2008/2009. Al-

though fuzzy keyword search seems to be of signif-

icant importance for user’s comfort, it only appears

sporadically in articles about SE schemes, mainly due

to the lack of efﬁcient implementations. See also Fig-

ure 3 for a graphical depictions of the trends over

time.

We also researched cryptographic technologies

and identiﬁed identity based encrpytion (IBE), at-

tribute based encryption (ABE), and bilinear maps

(BM) as the dominating asymmetric techniques.

However, from the corpus we could not determine a

speciﬁc future trend for any of these techniques.

4 CONCLUSION AND FURTHER

WORK

We initially started this project in order to ﬁnd a spe-

ciﬁc searchable encryption scheme for a particular use

case, but quickly became aware of the vast number of

existing schemes, together with the lack of any cur-

rent taxonomy of these schemes.

But usually one cannot decide on a particular

scheme without being aware of the number of users

and their read/write access privilege, resulting cryp-

tographic primitives and the support of efﬁcient and

secure updates.

All the categorization domains we described in

this paper are of signiﬁcant importance for choosing

the SE scheme for any use case.

We are currently in the process of implementing a

framework that allows for fast and usable experimen-

tation with different searchable encryption schemes

in real-world scenarios (Hab

ock et al., 2018). In ad-

dition we try to ﬁnalize our categorisation of papers

up until the current date, and try to identify future re-

search trends from this.

The quite recently published ﬁle injection attacks

on dynamic SE schemes and the deﬁnition of for-

ward and backward security properties in the context

Search, Find and Resolve: Towards a Taxonomy for Searchable Encryption Schemes

419

Figure 3: Supported Search Techniques 2003-2016.

of searchable encryption schemes evoke new chal-

lenges. be it by the need for thorough re-evaluation

and re-design of already existing schemes or by de-

signing new schemes with resilient security guaran-

tees in real-world environments.

ACKNOWLEDGEMENTS

Part of our work on the Searchitect project is funded

by Netidee

(Internet Privatstiftung Austria), grant

no. 2099.

REFERENCES

Boneh, D., Crescenzo, G. D., Ostrovsky, R., and Per-

siano, G. (2004). Public Key Encryption with Key-

word Search. In EUROCRYPT 2004, volume 3027 of

LNCS, pages 506–522. IACR.

Boneh, D., Sahai, A., and Waters, B. (2011). Functional

Encryption: Deﬁnitions and Challenges. In Ishai,

Y., editor, Theory of Cryptography: 8th Theory of

Cryptography Conference, TCC 2011, Providence, RI,

USA, March 28-30, 2011. Proceedings, pages 253–

273. Springer Berlin Heidelberg, Berlin, Heidelberg.

https://www.netidee.at

osch, C., Hartel, P., Jonker, W., and Peter, A. (2014).

A Survey of Provably Secure Searchable Encryption.

ACM Computing Surveys, 47(2):1–51.

Bost, R. (2016). Sophos - Forward Secure Searchable En-

cryption. Published: Cryptology ePrint Archive, Re-

port 2016/728.

Bost, R., Fouque, P.-A., and Pointcheval, D. (2016). Veriﬁ-

able Dynamic Symmetric Searchable Encryption: Op-

timality and Forward Security. Published: Cryptology

ePrint Archive, Report 2016/062.

Cash, D., Grubbs, P., Perry, J., and Ristenpart, T. (2015).

Leakage-Abuse Attacks Against Searchable Encryp-

tion. In Proceedings of the 22nd ACM SIGSAC Con-

ference on Computer and Communications Security,

pages 668–679, Denver, Colorado, USA. ACM.

Chang, Y.-c. and Mitzenmacher, M. (2005). Privacy

Preserving Keyword Searches on Remote Encrypted

Data. In ACNS 2005, volume 3531 of LNCS, pages

442–455. Springer Berlin Heidelberg.

Cui, S., Asghar, M., Galbraith, S., and Russello, G. (2017).

Secure and Practical Searchable Encryption: A Po-

sition Paper. In Pieprzyk, J. and Suriadi, S., edi-

tors, Information Security and Privacy: 22nd Aus-

tralasian Conference, ACISP 2017, Auckland, New

Zealand, July 5, 2017, Proceedings, Part I, pages 266–

281. Springer International Publishing, Cham. DOI:

10.1007/978-3-319-60055-0 14.

Hab

ock, U., Koschuch, M., Kramer, I., Schmidt, S., and

Tausig, M. (2018). Searchitect - A developer frame-

work for hybrid searchable encryption (position pa-

per). In Proceedings of the 3rd International Con-

ference on Internet of Things, Big Data and Security,

IoTBDS 2018, Funchal, Madeira, Portugal, March

19-21, 2018., pages 291–298.

Islam, M. S., Kuzu, M., and Kantarcioglu, M. (2012).

IoTBDS 2019 - 4th International Conference on Internet of Things, Big Data and Security

420

Access Pattern disclosure on Searchable Encryption:

Ramiﬁcation, Attack and Mitigation. In NDSS.

Liesdonk, P. V., Sedghi, S., Doumen, J., Hartel, P., and

Jonker, W. (2010). Computationally Efﬁcient Search-

able Symmetric Encryption. In SIAM Conference on

Data Mining SDM 2010, volume 6358 of LNCS, pages

87–100. Springer.

Liu, C., Zhu, L., Wang, M., and Tan, Y.-a. (2013). Search

Pattern Leakage in Searchable Encryption: Attacks

and New Constructions. IACR Cryptology ePrint

Archive.

M. Li, S. Yu, K. Ren, W. Lou, and Y. T. Hou (2013). Toward

Privacy-Assured and Searchable Cloud Data Storage

Services. IEEE Network, 27(4):56–62.

Naveed, M., Kamara, S., and Wright, C. V. (2015). In-

ference Attacks on Property-Preserving Encrypted

Databases. In Proceedings of the 22nd ACM SIGSAC

Conference on Computer and Communications Secu-

rity, pages 644–655, Denver, Colorado, USA. ACM.

Sedghi, S., Doumen, J., Hartel, P., and Jonker, W. (2008).

Towards an Information Theoretic Analysis of Search-

able Encryption. In Chen, L., Ryan, M. D., and

Wang, G., editors, Information and Communications

Security: 10th International Conference, ICICS 2008

Birmingham, UK, October 20 - 22, 2008 Proceedings,

pages 345–360. Springer Berlin Heidelberg, Berlin,

Heidelberg.

Song, D. X., Wagner, D., and Perrig, A. (2000). Practical

techniques for searches on encrypted data. In S&P

2000, pages 44–55. IEEE.

Stefanov, E., Papamanthou, C., Shi, E., and Encryption,

S. (2013). Practical Dynamic Searchable Encryption

with Small Leakage.

Tang, Q. (2012). Search in Encrypted Data: Theoretical

Models and Practical Applications. IACR Cryptology

ePrint Archive, 2012(648).

Wang, C., Cao, N., Li, J., Ren, K., and Lou, W. (2010). Se-

cure Ranked Keyword Search over Encrypted Cloud

Data. 2010 IEEE 30th International Conference on

Distributed Computing Systems, pages 253–262.

Zhang, Y., Katz, J., and Papamanthou, C. (2016). All

Your Queries Are Belong to Us: The Power of File-

Injection Attacks on Searchable Encryption. IACR

Cryptology ePrint Archive, 2016:172.

Search, Find and Resolve: Towards a Taxonomy for Searchable Encryption Schemes

421