ElGamal cryptosystem, multi-stage separation query processing (MSQP) protocol
1 Introduction
When multiple database owner wishes to collaborate using their individual cloud database by creating a distributed cloud database, you may need to create and maintain a database with very large distributed cloud privacy protection, but not included in his personal share confidential data in the database.
For example, a large number of records can be stored in a distributed database , the database combines data from different hospitals and even several different countries . This type of database can be managed by the World Health Organization, and can be huge. Patients without legal permission, may prevent different hospitals reallocate its data store. In the case of the database owner in different countries, some countries may have made laws requiring the patient confidential data stored in the actual data center located in the same country. Other countries may consider introducing a similar privacy laws in the future. For large cloud databases contain different companies the confidence of business data, there may be a similar situation. Again, this might not want to cooperate for the benefit of its customers to share information, several online trading site. Combined with patient data from different regions to analyze large data sets capability can improve the results of statistical analysis of medical data (see [3] , [4] ). Importantly, it must publish the answers to some of the aggregate queries throughout the very large distributed cloud database, especially as statistical information can help improve the quality of patient treatment and prediction of disease (see [5] ). Database owners may need to answer customer queries summary statistics presented in some way, like all records from all the databases are included in the same single public databases. However, the law requires that the database owner is not allowed to transfer their patient records to other entities. Therefore, we must protect the privacy of confidential data. Meanwhile, due to the distributed database may be very large, so that the communication cost must be minimized during the calculation.
Using multistage separation query processing (MSQP) protocol, which uses a split key of the homomorphic encryption
Statistical inquiry novel MSQP private treatment protocol. Participants MSQP agreement is the database owner, they represent the official registration entity, and there is no anonymous participants. Therefore, it is natural to assume that the database owner to be honest. For example, the database owner may be different hospitals, this is absolutely honest, they may not even curious. This is why MSQP agreement, we can put all the database owner deemed honest but curious participants, because in any case, because of legal requirements, must protect the confidentiality of data, in order to prevent the curiosity of all participants .
7.1 . Having a system for cryptographic protocol of splitting keys MSQP
In MSQP protocol, the database owner using the homomorphic encryption by the auxiliary security combined with calculated results. MSQP can use the properties specified in the following (P1) and (P2), any cryptographic system. This section describes only the general symbols and attributes may be used to describe MSQP cryptography protocol. To meet the required performance (P1) and two examples (P2) of the cryptographic system is given in Section ElGamal encryption scheme 5 preliminary round , ElGamal cryptosystem over six portions .
MSQP protocol cryptosystem having a public key pk all participants know the secret key and split Ski , i∈ [1: D] , where : [1 d] the set of all integers represented by 1 to d . In MSQP the agreement, the secret decryption key ski only database owner knows Di . All expressed as a set of secret keys SK . Public key pk just need to know all of the database owner; customers or outsiders should not be told. Accordingly, when the selection of a particular cryptosystem, the database owner can use secure encrypted communication between them to share a common key. This allows the database owner can easily share a common key between each other.
For the text T , which is expressed as the encryption E (t, PK) . Given ciphertext C , using the key to decrypt ski denoted by D (c, ski) . Decryption c by successively applying all split key to perform the following operations:
The following properties crucial cryptography.
-
(P1)
attribute split key.
Using the secret key to decrypt all be split must be properly restore the original plaintext
-
(P2) Generalized homomorphic properties. For any k plaintext 
, a given product 
decryption algorithm (1) must be restored plaintext sum:
In this case, we can also say that the password system meets the generalized additive homomorphic properties.
7.2 . Stage MSQP agreement
MSQP protocol using a query server asked . In all stages of the protocol in FIG 2 is described . In the data flow in FIG. 3 is shown .
In the following section 7.5 provides a general description of the algorithm MSQP protocol steps 1 section . 1 algorithm provides more detailed information on the application protocol steps of the MEC.
( 2 FIG . Phase MSQP protocol. )
( FIG. 3 . MSQP protocol data flow diagram for calculating and returning query results )
7.3 . Query sample and record the agreement in MSQP
主要功能的选择 F当客户端通过指示请求哪个统计信息作为查询样本的输出传达查询时，进行查询。对于众所周知的统计数据，用户足以表明该术语。另一方面，用户也很容易提交数学表达式作为要在查询输出中使用的公式。
通过指定条件为查询选择的记录集称为查询样本或查询集。属于查询样本的记录也称为样本记录。为了选择查询样本，客户端可以使用各种逻辑运算符，通过指定要保留的属性不等式来组合记录中属性的条件，并使用布尔运算符来组合这些条件。然后，数据库所有者可以将条件应用于其数据库部分，以选择查询样本中的记录部分。
假设每条记录a的分布式数据库包含n属性，在哪里n是一个正整数。我们假定所有属性都是数字，因为分类属性也可以编码为数字。表示记录中存储的属性a通过
要选择分布式数据库中的记录样本，每个查询都使用布尔表达式。表示为
所有布尔表达式的集合。可以通过以下两个规则来归纳定义此集合。
表2。数据库所有者，子样本及其记录。
Throughout the process, we consider using query Boolean expressions 
. It selects the records stored in the database. It is expressed as S = B (D) query sample, i.e. throughout the distributed database in the set of all records satisfying the condition of B. Let m = | S | is a record number of samples in the trumpet.