Method for managing a set of data packet compression rules

Abstract

Method for managing a set of data packet compression rules. The invention relates to a method for managing a set (RS) of compression rules applicable to data packets, called the initial set, characterized in that it comprises the following steps: A step of determining (S1) a set of rules, called the effective rules, for a batch (L) of data packets, called the given batch; A step of obtaining (S2) a reduced set (RRS) of compression rules by removing from the initial set (RS) compression rules that do not belong to the effective set of rules for the given batch (L). Figure 1

Description

Title of the invention: Method for managing a set of data packet compression rules technical field

[0001] The technical field is that of data compression and decompression.

[0002] More specifically, the invention relates to a method for managing a set of compression rules for data packets. These data packets are intended to be transmitted over a computer or telecommunications network. The method according to the invention aims in particular to reduce the size of the set of compression rules.

[0003] Modern networked computing and digital telecommunications operate on the principle of exchanging data packets between senders and receivers using protocols such as the IP protocol (acronym for Internet Protocol). A set of data packets relating to a communication is sent by the sender to a receiver. The packets contain the receiver's address so that they can be addressed using various routing algorithms, and the content of the communication is reconstructed by the receiver by regrouping the data contained in the different packets. To achieve complete communication, numerous protocols must be used, each fulfilling a role from the physical transmission of bits, the atomic elements constituting digital data, to the interpretation of data for computer applications. Bits are conventionally grouped into bytes, sets of eight bits.The various protocols are classically viewed as being organized in layers, with a protocol at one layer relying on a protocol at a lower layer. The OSI (Open Systems Interconnection) model and the Internet model describe the articulation of the various protocols between different layers.

[0004] For example, in a simplified version, the HTTP (Hypertext Transfer Protocol) protocol is an Application layer protocol that allows data to be transferred to or from a website. The data to be transferred is organized into data packets. The HTTP protocol uses the TCP (Transmission Control Protocol) protocol, which belongs to the Transport layer, for example, to request the transfer of data packets representing the content of a website to a user. The TCP protocol, for example, operates in three phases: establishing a connection, transferring data, and terminating the connection. Transport layer protocols will use a Network layer protocol, generally the IP (Internet Protocol) protocol in one of its IPv4 or IPv6 versions. The protocol is only concerned with the transmission of data organized into packets from one address to another, without dealing with establishing the connection or the order in which the data packets arrive within a transmission; these considerations are handled by the protocol of the higher Transport layer. The IP protocol will use a protocol of the Data Link layer, for example, the Ethernet protocol, which will organize the sending of the elementary data that constitutes the packet. This elementary data is bits, transmitted using the Physical layer, either via fiber optic cable or wireless communication, depending on the situation.

[0005] Other Application layer protocols will use other protocols from lower layers to perform their functions. For example, the CoAP (Constrained Application Protocol) protocol used in the Internet of Things generally relies on the UDP (UserDatagram Protocol) protocol of the Transport layer.

[0006] A common feature of all protocols across all layers is the organization of exchanged data into data packets. A data packet is organized into two parts: the first part is the header, which contains the information necessary for the protocol responsible for processing the packet; the second part is the payload, which contains the data carried within the packet itself. If packets of a given protocol contain one or more encapsulated packets of another protocol, the payload of a packet of the first protocol may contain the header of that other protocol. At the Data Link layer, the term frame is used rather than packet, and, in addition to the header, there is a footer indicating the end of the frame.

[0007] Packet headers include fields. A field is a continuous set of bits of fixed or variable length. When a field is variable in length, mechanisms exist to determine the actual length of the transmitted field. When the value in a fixed-length field is variable in length, padding bits are used to pad the value up to the field's fixed length. A field has an identifier that indicates the nature of the value it contains. The values ​​in the field are used by the relevant protocol to process the packet appropriately. For example, in the header of an IP packet, a field with the identifier Version will contain the version number of the protocol used. This field is four bits long. Other fields include, for example, the source and destination IP addresses of the packet.A header is therefore naturally divided into several sections, namely the header fields. The packet payload can also be divided into several sections, for example by defining sections of a constant length expressed in number of bits.

[0008] A packet analysis tool such as Wireshark or tcpdump allows the capture and analysis of data packets. This analysis separates a data packet exchanged over a network into its header and payload, reads the information contained in the header and the data contained in the payload, and identifies the various sections of a packet, both in its header and its payload. It should be noted that, following the encapsulation of data packets, the typical header of a packet will consist of several concatenated headers, one for each encapsulated protocol. For example, an Ethernet frame encapsulating an IP packet will begin with the IP header, followed by the UDP header. A UDP datagram carried by an Ethernet frame will then successively present the Ethernet header, followed by the IP header, and then the UDP header.

[0009] There is a clear need for data compression in computing and digital telecommunications to optimize the use of hardware and energy resources consumed by computer or telecommunications networks. This need explains the creation of protocols dedicated to compressing exchanged data packets. Examples of compression protocols include ROHC (Robust Header Compression), defined by the IETF (Internet Engineering Task Force) standardization body in RFC 3095 (Request For Comments, a specification document underlying the Internet), and SCHC (Static Context Header Compression), defined by the IETF in RFC 8724.

[0010] These protocols operate using compression rules that apply or not to a particular data packet depending on the values ​​taken by the different sections of that given packet. The compression rules used in the ROHC or SCHC protocols, or in other similar protocols, will, for example, attempt to erase a known value taken by a given section of a packet, and a complementary decompression rule will restore the erased value upon reception of the packet. Other rules may replace a given value with the index of a dictionary or may only erase a prefix of the value represented as a sequence of bits, a prefix that will be restored upon decompression. This results in very efficient compression of data packets, even if the compression is limited to the headers.

[0011] In all cases, the operation of data packet compression protocols will rely on the existence of sets of compression rules that describe the compression operations to be performed on header field values ​​or, more generally, on sections of data packets, provided that these values ​​meet conditions that trigger the application of the compression operations present in a rule. A compression rule is said to apply to a packet When all the compression operations present in the rule can be applied to the values ​​taken by the fields of the packet in question. For a given packet, there exists, within the set of compression rules, one or more rules that achieve the most efficient compression when applied to that particular packet. The most efficient compression is the greatest compression achieved by the different rules that can be applied to the packet in question. These particular rules can be called the optimal rules for the packet. In general, the compression protocol will seek to apply, for each data packet it must compress, a rule that is optimal for that packet in order to achieve the best compression for that given packet.Operations that do not perform any compression but potentially apply to all possible values ​​can be defined for completion reasons, so that it is possible, for any given package, to find a compression rule that applies to that package, even if no compression is performed.

[0012] All of these elements are well known in the prior art and will not be described further. State of the art

[0013] In general, compression rule sets are defined by experts based on their knowledge and an analysis of packets representative of a given traffic pattern. The idea that compression rule sets can be obtained through machine learning is beginning to emerge in the state of the art. Obtaining a compression rule set through machine learning would also start from a set of packets representative of the expected traffic, which would then be classically referred to as the training set.

[0014] To ensure the best possible operation of compression protocols, the quality of compression rule sets is a major issue. This quality is judged by the compression ratio achieved by a given rule set when it is deployed to run on a given traffic stream. The quality of a compression rule set can therefore be measured objectively by comparing the compression ratios achieved by different rule sets on the same traffic stream.

[0015] Another quality criterion for a compression rule set is its size. The smaller the set, for the same compression ratio, the better, because it will take up less memory space in the receiving and transmitting equipment. Furthermore, when the compression rule set is small, the compression protocol takes less time to determine which rule to apply. Finally, when a rule is used, its identifier must generally be transmitted with the compressed packet. to allow the reverse operation of decompression to be performed upon receipt of the packet. The more rules there are, the larger the size of the rule identifier used will be, and therefore the greater the number of bits used to encode this information in the transmitted packet, which goes against the objective of compression.

[0016] The two criteria of size and compression ratio are therefore used to judge the quality of a set of compression rules, whether this set is provided by an expert or obtained through machine learning. The size criterion is particularly important in the context of connected objects, which have limited memory and for which a maximum size for the set of compression rules may be imposed. The management method according to the invention aims to reduce the size of a given set of compression rules until it reaches a specified size while maintaining good performance with regard to the compression ratio.

[0017] In the prior art, compression rule sets are often determined by experts. If this set must have a maximum size, the expert will then ensure that this size is not exceeded during the rule set construction. Such a process is therefore performed manually and prevents the use of compression rule sets that adapt to traffic, since changing a compression rule set requires expert intervention.

[0018] In cases where the compression rule set is obtained through machine learning, the latter will generally seek to produce compression rules adapted to the training set being fed into the machine learning algorithm and will tend to add increasingly more suitable rules. The compression rule sets obtained through machine learning are therefore generally very large, which makes them impractical.

[0019] The invention improves the situation. Description of the invention

[0020] According to a first functional aspect, the invention relates to a method for managing a set of compression rules applying to data packets, referred to as the initial set, characterized in that it comprises the following steps: • A step of determining a set of compression rules, called efficient rules, for a batch of data packets, called a given batch; • A step of obtaining a reduced set of compression rules by removing from the initial set the compression rules that do not belong to the effective set of rules for the given batch.

[0021] Thanks to the invention, a reduction in the size of the initial set of compression rules is achieved. In this way, an initial set that was of a size If the size of a packet is too large to be deployed in a connected device or used by telecommunications equipment, it will be reduced. This size reduction takes into account a given batch of data packets, which are representative of the expected traffic that must be compressed by the set of compression rules. The reduction is achieved by first targeting the rules that are not efficient at compressing the packets belonging to this given batch.

[0022] The advantage of the invention is therefore to reduce the size of the compression rule set by removing rules from this set, while ensuring that the effective rules for the packets in a given batch, packets representative of the expected traffic, are retained. The reduction in the size of the compression rule set is thus achieved while preserving its quality in terms of the compression performed by the retained rules.

[0023] According to one embodiment of this functional aspect of the invention, the method further includes a step of supplying the reduced set of compression rules.

[0024] After this reduction, the reduced set of compression rules can be provided for use in performing packet compression according to a compression protocol.

[0025] According to another embodiment of this functional aspect of the invention, a rule is determined as belonging to the set of effective rules when there is a packet in the given batch for which said compression rule achieves the greatest compression of the packet among the compression rules of the initial set.

[0026] Thanks to this embodiment, an objective criterion is defined that makes it possible to determine the effective rules to be retained for a given batch of packets. The rules considered effective are those that are optimal for at least one data packet belonging to the given batch. A rule is optimal for a packet when it achieves the greatest compression of that packet compared to the compressions that can be achieved among all the compression rules in the set. In general, an optimal rule for a packet will be the one chosen by the compression protocol to be applied to it. By retaining the optimal rules from among the effective rules that are kept by the management process, the latter will disrupt the operation of the compression protocol as little as possible since it retains the effective rules that will be most frequently used.

[0027] According to another embodiment, which may be implemented alternatively with the previous embodiment, a rule is determined to belong to the effective rule set when said rule achieves compression above a given threshold of the packets in the given batch to which said rule applies.

[0028] This embodiment provides an alternative way to determine which effective rules to retain. A rule is considered effective here when it achieves compression exceeding a given threshold for the packets to which it applies. In this embodiment, the rules achieving the greatest compression are retained.

[0029] According to another embodiment of this functional aspect of the invention, which may be implemented alternatively or cumulatively with the preceding embodiments, the method further comprises the following steps: • A step of obtaining a compression penalty caused by the deletion of a rule from the reduced set; • A step of removing a rule from the reduced set that minimizes the penalty obtained.

[0030] If simply removing rules that are ineffective or only slightly effective is insufficient to adequately reduce the rule set, the two steps described above are performed by the method to further reduce the size while maintaining good compression quality for the rule set relative to the given batch. To achieve this, a penalty is applied to assess the effect of removing a rule from the reduced set. Indeed, removing a given rule indicates that some packets in the expected traffic will no longer be compressed by that rule. A simple way to calculate the penalty is to count the bits erased by a rule in a packet when it is applied to and compresses that packet.

[0031] A rule is then removed, which is the one for which the penalty is lowest. In case of a tie, the removed rule can be randomly selected from among those for which the penalty is lowest, or chosen according to the order in which the rule appears in the reduced set. The succession of these two operations ensures that the size of the reduced set of compression rules decreases by at least one rule.

[0032] According to another embodiment, which can be implemented cumulatively with the previous embodiment, a size of a set of compression rules, called the target size, is provided and the method includes an iteration of the steps of obtaining a compression penalty and deleting a compression rule until the size of the reduced set is less than the target size.

[0033] In this embodiment, the sequence of two operations—obtaining a penalty and deleting a rule to minimize the penalty—is repeated until a target size is reached. In some embodiments, the reduced set can then be provided for use in compressing data packets as expected.

[0034] According to another embodiment, which can be implemented cumulatively with the previous embodiment, the step of obtaining a compression penalty takes place Note that a compression rule removed from the reduced set will be replaced by another rule from the reduced set to perform data packet compression.

[0035] Thanks to this embodiment, the compression penalty that guides the removal of a rule from the reduced set is made more precise. Indeed, if a rule is removed from the reduced set, this does not mean that the packets that would have been compressed by that rule will no longer be compressed at all. They will generally be compressed by another rule, which may be less efficient. The penalty can therefore be obtained as the difference between the compression performed by the removed rule and the compression performed by the rule that will be chosen by the compression protocol to compress packets in place of the removed rule.

[0036] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, the step of obtaining a compression penalty further takes into account that several compression rules may be removed from the reduced set and replaced by another rule, present in the initial set and absent from the reduced set, and that the removal step then includes the removal of said several rules and the addition of said other rule if this removal combined with this addition minimizes the penalty obtained during the obtaining step.

[0037] Thanks to this embodiment, the compression penalty that guides rule deletions is further refined. Indeed, at some point, a rule may have been deleted during previous steps of the process and therefore be absent from the reduced set, but it can nevertheless provide an interesting replacement for a rule to be deleted. Simply put, if a rule that had been deleted is reintroduced into the reduced set, at least two rules must be deleted to ensure that the number of rules in the reduced set always decreases by one. The penalty therefore takes into account the possibility of deleting several rules from the reduced set and reintroducing a rule that is no longer in the reduced set but was in the initial set in order to replace the deleted rules when the compression protocol uses the reduced set of rules to compress data packets.The penalty is calculated here by subtracting the penalty of the added rule from the sum of the penalties of the several removed rules.

[0038] According to another embodiment of the invention, which may be implemented alternatively or cumulatively with the preceding embodiments, the initial set of compression rules is represented by a data structure in which a compression rule, referred to as the first rule, is linked to another compression rule, said second rule, when, if the first rule applies to a given packet, the second rule applies to the same packet.

[0039] Thanks to this embodiment, the execution of the process is simplified. Having such a data structure makes it possible to quickly determine whether a rule is more generic than a more specialized one, that is, whether it can be applied to all data packets to which the latter can be applied. This information can be used to obtain compression penalties because it indicates which rule can replace a deleted rule. Obtaining this information directly, since it is stored in the data structure itself, greatly accelerates and simplifies the implementation of the process according to the invention.

[0040] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, obtaining a compression penalty takes into account the number of packets in the given batch for which the deleted rule(s) achieve the greatest compression among the rules in the reduced set.

[0041] This embodiment further refines the penalty that guides the removal of rules from the reduced set. To this end, the penalty takes into account, within the given batch, which is representative of the expected traffic, the packets for which the rules to be removed achieve the greatest compression and will therefore be chosen by the compression protocol to compress the packets of the expected traffic. A rule that achieves the greatest compression for a given packet among a set of possible rules is said to be optimal for that packet. If a rule compresses a packet by ten bits, but is optimal for only one packet in the given batch, the penalty can be ten bits; whereas a rule that compresses a packet by only one bit, but is optimal for one hundred packets in the given batch, will have a penalty of one hundred bits in total.This consideration of the number of packets for which the deleted rule(s) are optimal must also take into account that a deleted rule will be replaced by another rule from the reduced set to effectively compress these packets.

[0042] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, the size of a set of compression rules is the number of compression rules belonging to the set of compression rules.

[0043] This embodiment provides a simple way to calculate the size of a set of compression rules. Naturally, this size is the number of rules in the set. The process will therefore reduce the number of rules in the managed rule set until a target number is reached.

[0044] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, the size of a set of rules of compression is calculated by taking into account the number of compression rules belonging to the set of compression rules as well as the complexity of the rules belonging to the set.

[0045] It is possible to calculate the size in a more elaborate way, taking into account the complexity of the compression rules present in the compression rule set. In this case, a rule containing many more detailed compression operations will have a larger size than a simpler rule, and this difference will be reflected in the calculation of the size of a compression rule set. An example of the calculation can be given as follows: the size of a rule is the number of different effective compression operations present in the rule, and the size of the rule set is the sum of the sizes of the rules present in the set. More complex formulas can be used, taking into account the complexity of the compression operations.

[0046] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, the determination step includes calculating for a rule in the initial set a score obtained by multiplying the compression achieved by said rule by the number of packets in the given batch for which said rule achieves the greatest compression among the rules in the initial set.

[0047] Thanks to this embodiment, complete information on the compression capabilities of the rules in the initial set is obtained. The value of a compression rule depends on the compression it performs when applied to a packet, multiplied by the number of packets for which this rule achieves the greatest compression among all possible rules and will therefore likely be used for compression. Since this calculation is performed for the packets in the given batch, which is representative of the expected traffic, the score thus calculated provides an indication of the amount of compression a given rule will perform on expected traffic. This score also reveals which rules do not achieve the greatest compression for any packet in the given batch, as these rules will then have a score of zero.

[0048] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, the step of obtaining the reduced set includes the removal of the initial set of compression rules whose score is less than a given threshold.

[0049] This embodiment offers an alternative. Rather than directly deleting the rules that do not achieve the highest compression for any package, this embodiment deletes the rules whose score is below a given threshold. In this way, more rules are removed, with the risk of removing rules that are optimal for certain packages in the given batch.

[0050] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, a size of a set of compression rules, called the target size, is provided and the step of obtaining the reduced set includes the removal of the initial set of compression rules with the lowest score until the size of the reduced set is less than the target size.

[0051] This embodiment also offers another alternative implementation of the invention. Here, the rules are removed one after the other based on the calculated score until the target size is reached. The advantage of this embodiment is its simplicity, at the expense of accounting for the replacement of removed rules by other rules in the actual packet compression, a factor addressed in other embodiments.

[0052] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, the initial set of compression rules is obtained by a machine learning algorithm and the given batch was included in the training set used to obtain the initial set.

[0053] Thanks to this embodiment, the method according to the invention is used in the context of machine learning of compression rules. The method according to the invention thus solves a significant problem in machine learning, namely that it produces large sets of compression rules. After the machine learning phase, the method according to the invention automatically reduces the size of the resulting compression rule set to a target size in certain embodiments, for example, to allow deployment in telecommunications equipment that imposes a specific size on the deployed compression rule set.

[0054] To ensure that reducing the size of the compression rule set does not compromise the quality of the compressions performed by the reduced set, the method according to the invention is carried out by taking into account a given batch of packets. This given batch is representative of the expected traffic that will be compressed by the reduced set of compression rules. When the set to be reduced is derived from machine learning, the given batch may be the training set used for the machine learning or may be included within the training set. In this way, the machine learning and the reduction of the compression rule set use the same set of packets representative of the expected traffic and are consistent with each other.

[0055] The method according to the invention can be carried out after another method for managing a set of compression rules in which, after machine learning, a set of compression rules is checked against a set of packets that were not used during the machine learning, in order to ensure that the machine learning is not over-adapted to the training set. In this case, the overall acquisition of a set of compression rules is achieved through a succession of three distinct methods: • Machine learning of a set of compression rules by a machine learning process using a batch of data packets, representative of the expected traffic, as a training set; • Automatic control of the generalization of the compression rule set obtained by machine learning using another batch of data packets, separate from the training set, also representative of the expected traffic, as a control set; • Automatic reduction of the size of the compression rule set to adapt it to the constraints of the equipment where the set will be deployed, this reduction using at least the training set to take into account the expected traffic, and potentially the combination of the training set and the disjoint control set. This reduction is the object of the method according to the invention.

[0056] In one embodiment, the given batch used in the process according to the invention comprises, in addition to the training set, the other batch of packets, also representative of the expected traffic, used as a control set. In this way, a maximum number of packets representative of the expected traffic are used in the process according to the invention.

[0057] According to another embodiment of the invention, which may be implemented alternatively or cumulatively with the preceding embodiments, the data structure representing the initial set was calculated by the machine learning algorithm that enabled the initial set to be obtained.

[0058] A compression rule learning algorithm can operate by starting with general rules and searching for more specialized rules that apply to the data packets in the batch of packets used as the training set. This type of learning algorithm can produce, as a side effect, a data structure that retains information about which rule was created from which other rule, and more generally, which rule is more specialized than another. In this embodiment, this information, which was a side effect of the machine learning algorithm, becomes very useful for accelerating the management process according to the invention.

[0059] According to another embodiment, which may be implemented alternatively or cumulatively with the preceding embodiments, the compression rules of the initial set are compression rules according to the SCHC protocol.

[0060] Thanks to this embodiment, the process applies to the SCHC compression protocol and makes it possible to perform compression according to this protocol with small sets, suitable for Internet of Things equipment.

[0061] In other embodiments, the compression rules are those of other compression protocols.

[0062] According to a first material aspect, the invention relates to a management entity managing a set of compression rules applying to data packets, referred to as the initial set, comprising the following modules: • Module for determining a set of compression rules, called efficient rules, for a batch of data packets, called a given batch; • Module for obtaining a reduced set of compression rules by removing from the initial set compression rules that do not belong to the effective set of rules for the given batch.

[0063] According to a first embodiment of this first hardware aspect, the management entity further comprises the following modules: • Module for obtaining a compression penalty caused by the deletion of a rule from the reduced set; • Module for removing a rule from the reduced set that minimizes the penalty obtained.

[0064] In embodiments, the management entity further includes a module for supplying the reduced set.

[0065] According to another embodiment of this first material aspect, the invention relates to telecommunications equipment comprising a management entity according to the invention.

[0066] According to another embodiment, the invention relates to a mobile terminal comprising a management entity according to the invention.

[0067] According to another material aspect, the invention relates to a computer program, the program comprising code instructions which, when executed by a processor, carries out the steps of the management process defined above.

[0068] Finally, according to another material aspect, the invention relates to a data carrier on which is recorded a computer program comprising sequences of instructions for the implementation of the management process defined above.

[0069] Data carriers can be any entity or device capable of storing programs. For example, the carriers can include a storage means, such as a ROM, for example a CD-ROM or a circuit ROM microelectronics, or a magnetic recording medium such as a hard drive. Alternatively, the media can be transmissible, such as an electrical or optical signal, which can be transmitted via an electrical or optical cable, by radio, or by other means. The programs according to the invention can, in particular, be uploaded to a network such as the Internet. Alternatively, the information medium can be an integrated circuit in which the program is incorporated, the circuit being adapted to execute or to be used in the execution of the process in question. The program according to the invention can use any type of computer technology in terms of compiled or interpreted programming languages, or a combination of both, as well as in terms of operating systems. Brief description of the figures

[0070] The invention will be better understood upon reading the following description, given by way of example, and made with reference to the accompanying drawings in which:

[0071] [Fig-1] represents a management entity carrying out the process according to the invention.

[0072] [Fig.2] represents a series of steps carried out during the process according to the invention.

[0073] [Fig.3] represents an example of data structure used in the process according to the invention. Detailed description

[0074] The [Fig. 1] represents a management entity 100 capable of carrying out a management process according to the invention.

[0075] In the embodiment shown in [Fig. 1], the management entity 100 comprises the following modules: • A module 101 for determining a set of rules, called effective rules, for a batch L of data packets, called the given batch; • A module 102 for obtaining the reduced set RRS by removing from the initial set RS the rules that do not belong to the effective rule set for the given batch L; • A module 103 supplying the reduced RRS assembly; • A module 104 for calculating the compression penalty caused by the removal of a rule from the reduced RRS set; • A module 105 for removing a rule from the reduced RRS set that minimizes the calculated penalty.

[0076] In other embodiment examples, the management entity does not include a 103 module for supplying the reduced RRS set and can, for example, be reduced to modules 101 and 102.

[0077] Management entity 100 has the hardware architecture of a conventional computer. It includes, in particular, a processor, a random access memory of RAM type and read-only memory such as Flash type memory, ROM, (not shown in the figure) as well as input-output devices such as, in some cases, keyboards and / or screens (not shown in the figure), and network ports allowing communication with other entities and servers through a communication network such as the Internet not shown in the figure.

[0078] The management entity will, for example, be a computer server that performs the management process as part of the work of establishing sets of compression rules before their deployment to one or more devices that will subsequently apply the rules of the managed set to data packets that they process in the context of real traffic. The management entity can also be deployed in a cloud architecture in which the user will not have a dedicated server but will request computing, memory, and network resources that will perform the functions of the management entity.

[0079] In other embodiments, the management entity will itself be part of a device that applies compression rules. The management entity is then a component of the device responsible for improving, in successive steps, the set or sets of rules that the component uses to compress the traffic processed by the device.

[0080] Such equipment may be, for example, a router. This router may be deployed in local area networks (LANs), metropolitan area networks (MANs), or wide area networks (WANs). The equipment may also be a gateway providing a link between distinct networks, for example, a home gateway or an access gateway for Internet of Things (IoT) devices, or a gateway providing a link between distinct protocols, for example, a gateway translating Voice over IP (VoIP) communications to an IP core network or, conversely, to a telephone access network or a fixed telephone network, for example, a corporate network. The equipment may also be a base station used in a mobile telecommunications network.The equipment can also be a mobile terminal. In these different embodiments, management entity 100 reduces the size of an initial set of compression rules RS to obtain a reduced set RRS. In some embodiments, the process can ensure that the reduced set RRS is smaller than a given size S, referred to as the target size. In other embodiments, the reduction in size from the initial set RS to the reduced set RRS occurs without targeting a specific size.

[0081] The initial set of compression rules RS could be established by an expert or by a machine learning process. The RS set could, for example, be a set of compression rules according to the SCHC protocol and be adapted to a particular traffic, that is, adapted to compress data packets used in a given protocol context, for given equipment. The RS set could then be deployed for use in a piece of equipment, and the management entity, in this case, could be a component of the equipment in question that will apply the process to reduce its size to the target size S and provide the reduced set RRS.Alternatively, after obtaining a first version of the RS set of compression rules, it is transmitted to management entity 100, which is a server that will not use the RS set to perform compressions but is dedicated to reducing the size of the compression rule sets.

[0082] In one embodiment, the RS set was obtained by machine learning, and a batch L, called the given batch, is then, in this example, the set of data packets that were used as representative examples of the traffic to be compressed to perform this machine learning; that is, the training set. The notion of a training set is well known in the field of machine learning and is not detailed further. In a machine learning context, other packets may have been used as a control set to ensure that the RS set of compression rules is not too specialized. These other packets can be added to the training set to form the given batch L of data packets, which should be representative of the traffic expected to be compressed by the rules of the RS and RRS sets.

[0083] In a first SI step, visible in [Fig. 2] which will be described later, module 101 of management entity 100 determines a set of rules, called efficient rules, for the given batch L of data packets. The rules determined here from the set RS of compression rules are the most efficient rules for compressing the packets of batch L, and should therefore be retained, if possible, in the reduced set RRS.

[0084] One way to determine the efficient rules in the first step SI is to search for the rules in the initial set RS that are optimal for at least one packet in the given batch L, that is, those that achieve the best compression of a packet among the set of rules RS. Indeed, an optimal rule will very likely be chosen by the compression algorithm to compress the packet in question. Since the packets in the given batch L are representative of the expected traffic, it can be assumed that a rule that is not optimal for any packet in the given batch L will rarely be used and will therefore not be efficient. Such a rule can be removed from the set RS to obtain the reduced set RRS without too much impact the overall efficiency of the reduced RRS set of compression rules for compressing real traffic.

[0085] It can be noted that rules that are not optimal, that is, rules for which there is no packet in the given set L for which the rule achieves the greatest compression compared to the compressions achieved by the other rules in the RS set, may still be of interest because they could prove optimal for packets in real traffic that are not present in the given set L. The given set L aims to be representative of the traffic that the RS, RRS sets of compression rules will have to compress but cannot represent all possible examples of packets that may need to be compressed.

[0086] Another way to determine the efficient rules during the first SI step is to calculate a score for each rule by multiplying the compression achieved by the rule when applied to a packet by the number of packets in the given batch L for which the rule is optimal. This score corresponds to the compression achieved, that is, the number of bits erased, by a given rule on all the packets in the given batch L. Rules that are not optimal for any packet in the given batch L then have a score of zero.

[0087] It is then possible to determine in the first step IF the rules are effective by choosing those which have a score greater than a given threshold and this method joins the previous one by choosing a zero threshold.

[0088] In a second step S2, visible in [Fig.2] which will be described later, module 102 obtains a reduced set RRS by removing from the initial set RS the rules which were not determined during step SI as belonging to the set of effective rules.

[0089] The production of the reduced set RRS consists of removing from the initial set RS the compression rules which were not determined during the SI step as being effective according to a certain criterion.

[0090] In a third step S3, visible in [Fig. 2] which will be described later, the module 103 provides the reduced assembly RRS when the latter is smaller than a given size S, referred to as the target size. This supply step S3 is present only in certain embodiments. In some embodiments, the supply step S3 takes place when the reduced assembly RRS has reached a size smaller than a given size S, referred to as the target size.

[0091] This S3 provisioning step can be carried out in several embodiments. In one embodiment, the management entity 100 is part of a telecommunications device that uses the RS compression rule set to perform data packet compression of traffic carried by the device and seeks to reduce the size of the RS compression rule set to a target size S. In this case, module 103 will make the reduced RRS set available by replacing the compression rule set used by the equipment to compress data packets, which was initially the RS set, with the reduced RRS set. In one embodiment, module 103 can transfer a file containing the reduced RRS set, or any other computer object representing the reduced RRS set, to other entities that wish to use it. For example, telecommunications equipment could have transferred to management entity 100, for size reduction, an RS set of compression rules along with a given batch L of data packets and a target size S, and, after the process is carried out, module 103 provides the reduced RRS set in response.In another embodiment, management entity 100 has received a set RS obtained by machine learning using a given batch L of data packets as the training set, and management entity 100 will carry out the process according to the invention in order to reduce the size of the set RS of compression rules to a target size S to obtain the reduced set RRS. This reduced set RRS will then be used in place of the initial set RS.

[0092] In other embodiments, the management entity 100 reduces the size of the initial set RS into a reduced set RRS, and the latter is used by one or more telecommunications devices as the reduced set RRS is reduced, without the need for a delimited step of supplying the reduced set RRS. The reduced set RRS and the initial set RS are simply available continuously throughout the operation of the process.

[0093] In a fourth step S4, visible in [Fig.2] which will be described later, module 104 gets a compression penalty caused by the deletion of a rule from the reduced set RRS.

[0094] During step S2, when removing from the initial set RS the rules that were not determined to be effective during step SI is insufficient to obtain a sufficiently reduced set RRS, the process can continue by attempting to remove additional compression rules. This removal will be guided by a compression penalty obtained during step S4. In some embodiments, the objective of these removal operations is for the reduced set RRS to reach a target size S.

[0095] The step S4 of obtaining a compression penalty can be performed by carrying out calculations by module 104 or by receiving calculations performed by other entities not shown in [Fig. 1]. Rather than being calculated, the penalties can be estimated or approximated using other methods. Obtaining a compression penalty S4 can be carried out in a first mode of This is achieved by searching, for each potentially deleted compression rule, which packets the potentially deleted rule is optimal for and which rule would become optimal if the currently optimal rule were deleted. Once these packets are found, it is possible to calculate the difference in compression achieved between the deleted optimal rule and the new optimal rule. This difference in compression constitutes the compression penalty. The greater the penalty, the more deleting a rule and replacing it with another rule as the optimal rule for packets will result in a decrease in the compression of those packets. The calculation of the compression penalty takes into account the number of packets for which the deleted rule is optimal in order to calculate an effective penalty in bits when a deleted rule is replaced by another to effectively compress packets.

[0096] In some embodiments, the calculation of the compression penalty takes into account all the rules present in the initial set RS. Indeed, rather than deleting a single rule, it may be advantageous to delete several compression rules and reintroduce one that had been previously deleted, for example, one that had not been determined SI to be among the effective rules. The calculation of a compression penalty will then combine the compression losses due to the deletion of two rules with the gain provided by the addition of a rule that was previously removed from the reduced set RRS of compression rules.

[0097] The description in [Fig.3] presents a detailed example of the implementation of these embodiments.

[0098] In other, simpler embodiments, the compression penalty can be obtained solely by considering the compression scores achieved by the compression rules removed from the RS set, without taking into account the replacement of the removed rules by other compression rules. In these embodiments, the rules to be removed are determined by comparing scores attached to the compression rules, independently of the packages to which the rules apply. The advantage of this embodiment is the greater simplicity in obtaining the minimum compression penalties and determining the removed rule.

[0099] In a fifth step S5, visible in [Fig.2] which will be described later, module 105 removes a rule from the reduced set RRS which minimizes the calculated penalty S4.

[0100] Once the compression penalties have been obtained, the process will remove from the reduced RRS set a compression rule that minimizes the compression penalty caused by this removal. This removal may correspond to the simple removal of a single rule or to the combined removal of several rules. with the reintroduction of a rule present in the initial RS set that had previously been removed.

[0101] In some embodiments, the acquisition steps S4 and deletion steps S5 are iterated until the reduced set RRS reaches a target size S that has been provided. The reduced set RRS can then be provided in a step S3 in an example embodiment that includes this step.

[0102] Fig. 2, for its part, represents a series of steps carried out during the process according to a particular embodiment.

[0103] In the example of [Fig. 2], the embodiment begins with a step SI of determining efficient rules from among the initial set RS of compression rules for the given batch L. The initial set RS could be obtained by a machine learning algorithm using the given batch L as the training set. The objective of the method according to the invention is to obtain a reduced set RRS of compression rules starting from the set RS. The determination SI of the efficient rules ensures that the efficient rules are preserved to the maximum extent possible even when reducing the size of the initial set RS to that of the reduced set RRS. One possibility is to retain in the reduced set RRS the optimal compression rules, that is, the rules for which there exists a data packet in the given batch L for which the considered rule achieves the best compression among all the compression rules considered.

[0104] The embodiment continues with a step S2 of obtaining a reduced set RRS by removing from the set RS rules that were not determined SI to be part of the effective rules in the previous step SL. In the embodiment example in [Fig. 3], a target size S has been provided. If this step S2 produces a reduced set RRS of compression rules whose size is less than the target size S, then the process will continue with step S3, as shown by the test in [Fig. 2]. The size of a compression rule set can be simply measured by counting the number of compression rules present in the set under consideration. Other methods for defining the size of a compression rule set can be considered that take into account the number and complexity of the elementary compression operations that form the compression rules.

[0105] If the target size S is reached, the method according to the invention comprises, in this embodiment, a step S3 of providing the reduced set RRS of compression rules. In other embodiments, the reduced set RRS can be used at any time to perform data packet compression without requiring a provisioning step S3. The reduced set RRS can be provided by transferring a file or any other relevant computer object. from management entity 100 to a device that needs to use the RRS compression rule set. In this context, management entity 100 can be a server dedicated to managing and improving compression rule sets. The reduced RRS rule set can also be made available by providing access to requests to obtain the RRS rule set. In this context, management entity 100 can be a component of a telecommunications device that uses the reduced RRS compression rule set to compress the traffic it transmits. Management entity 100 then makes available to the telecommunications device that encompasses it an RRS version of the initial RS compression rule set, reduced to the target size S.

[0106] If the target size S is not reached, the process, in this embodiment, includes a step S4 for obtaining a compression penalty caused by the deletion of a rule from the reduced set RRS. Since the rules remaining in the reduced set RRS after the deletion step S2 are effective rules for the packets in the given batch L, the deletion of a rule from the reduced set will impact the compression efficiency of the rules in the reduced set RRS. Therefore, compression penalties associated with rule deletions are obtained. Examples of penalty calculations are given in the description of [Fig. 3]. The compression penalties can take into account that only one rule is deleted, but also that a previously deleted rule can be reintroduced.In this case, two rules must be removed to compensate for the reintroduction of a rule and ensure a reduction in the size of the reduced set of compression rules.

[0107] The embodiment continues with a step S5 of effectively deleting at least one rule from the reduced set RRS of compression rules. In this embodiment, if the target size S is reached, the process will continue with step S3 of supplying the reduced set RRS. Otherwise, a further sequence of steps S4-S5 may take place.

[0108] Fig. 3, meanwhile, represents a data structure used in certain embodiments of the process.

[0109] Figure 3 shows an example of a data structure used in certain embodiments of the method of the invention. A set RS of nine compression rules, R1 to R9, is found in our example data structure. These nine rules, R1-R9, are organized in a data structure that is a tree. Rule R1 has two descendant rules, R2 and R3. Rule R2 has two descendant rules, R4 and R5. Rule R3 has one descendant rule, R6. Rule R4 has two descendant rules, R7 and R8. Rule R8 has one descendant rule, R9. Rules R6, R7, and R9 have no descendant rules. We will explain later under what A rule is defined as either a descendant or a parent of another. Such a data structure can be obtained through machine learning of the RS compression rule set.

[0110] The R1-R9 rules of the RS set are rules designed to compress data packets whose header includes five fields Fl to F5. The compression rules consist of compression operations that can be applied to each of the Fl to F5 fields of the packets targeted by the compression rules. Each rule RI to R9 therefore comprises five compression operations labeled Fl to F5. These compression operations can be of three types: • An EQUL operation will be applied to packets where the field targeted by the operation takes exactly the value specified in the operation. In this case, the value specified by the operation is not sent. This value will be reconstructed upon receipt of the compressed packet. • A PREF operation will be applied to packets where the field affected by the operation takes a value, expressed as a bit sequence, whose prefix has the value specified by the operation. In this case, the prefix specified by the operation is not sent. Again, the value will be reconstructed upon receipt of the compressed packet. • An IGNR operation will ignore the field and apply it to all possible values, and will not perform any compression. This type of operation ensures that all packets can be associated with a compression rule, even if that rule does not perform any compression.

[0111] These three compression operations are inspired by the compression operations defined in the SCHC protocol, knowing that, in this protocol, a fourth type of operation exists in which values ​​of a field are replaced by the indices of a dictionary if they are equal to values ​​present in this dictionary.

[0112] Let us consider rule R4, which presents the three types of operation described above, and let us consider a packet whose fields Fl to F5 take the following values: {Fl = 11010101; F2 = 01001000; F3 = 11; F4 = 11; F5 = 1}. Compression rule R4 can indeed be applied to this packet, and the effect of applying the compression operations would be as follows: • For the Fl field, the prefix 110101 would not be transmitted. The last two bits of the value taken by the Fl field in the packet would be transmitted, namely 01. • For field F2, the value 01001000 as a whole would not be transmitted. • For field F3, the prefix 1 would not be transmitted and only the second bit, also equal to 1, would be transmitted. • For field F4, the value 11 would not be transmitted. • For field F5, the operation ignores the field and transmits the value unchanged 1.

[0113] Finally, the packet transmitted after application of the R4 compression rule would be as follows: {F1 = 01; F2 = _; F3 = 1; F4 = _; F5 = 1}, where "_" indicates that no value is transmitted for a given field (here, fields F2 and F4). Along with the compressed packet, the R4 rule identifier would be transmitted to allow the receiving entity to reconstruct the original value.

[0114] In this example, the five fields of the initial packet formed a total of 21 bits. The finally transmitted packet comprised a total of 4 bits. Applying compression rule R4 therefore saves 17 bits during packet transmission (not taking into account that the R4 rule identifier must also be transmitted). This number of bits saved will always be the same for any packet to which compression rule R4 can be applied. This is the compression rule score, which is displayed for each rule in [Fig. 3]. This score ranges from 2 for rule RI, which includes only one prefix deletion rule 11 for the Fl field, to a score of 21 for rule R6, which deletes all fields from Fl to F5 in a data packet as soon as they take the values ​​specified in the compression operations.

[0115] The RI to R9 rules are organized according to a specialization relationship. When a rule is a parent of another, it is less specialized and will apply to all packages to which the descendant rule can apply. This specialization relationship can be determined by considering only the operations that constitute the compression rules.

[0116] During machine learning of a set of RS compression rules, such a data structure representing the specialization relationship between rules in the set can be obtained as a side effect of the learning process. Indeed, learning can be done starting from a very unspecialized rule, for example, a rule that ignores the values ​​of all fields and passes them unchanged, and then applying operational modifications that will increasingly specialize the rules.

[0117] In the example shown in [Fig. 3], the given set L, which may have been used as a training set in a machine learning algorithm, comprises nine packets pl to p9. To each rule RI to R9 of the set RS are attached the packets among the given set L to which the rule applies. For example, rule RI applies to all packets pl to p9 of the given set L, whereas rule R7 does not only applies to the two packets p2 and p3, and rule R9 only applies to packet p8. Packets pl to p9 are not further specified.

[0118] A final indication presented in [Fig. 3] concerns the packets for which a given rule achieves the greatest or best compression among all compression rules. These packets are then boxed in the list of packets associated with a given rule. For example, rule RI is optimal for packet p9, and therefore applies to this packet, and also applies to all packets p1 to p8. Similarly, rule R7 applies to packets p7 and p8 and is optimal for packet p7. Rule R2 applies to packets p2, p3, p6, p7, and p8 but is not optimal for any of these packets. Rule R9 applies only to packet p8 and is optimal for it, and so on. Often, a compression protocol will choose the optimal rule for a given packet, that is, the one that achieves the best compression for that given packet among all available compression rules.This justifies the importance of having this information, which can also be obtained through machine learning of the RS set of compression rules.

[0119] We will now describe an example of an embodiment of the process according to the invention. Other embodiments are possible.

[0120] A first step is to determine the efficient rules for the batch L of data packets pl to p9 from among the rules RI to R9. In the example we will present, the rules chosen as efficient are those that achieve the best compression for at least one of the packets pl to p9 of the given batch L. In the example in [Fig.3], all the rules are efficient according to this criterion except for rules R2 and R4, which do not achieve the best compression for any packet of batch L among the rules RI to R9 of the set of rules RS.

[0121] In our example, these two rules R2 and R4 are therefore removed from the initial set RS to form the reduced set RRS. This therefore includes the rules RI, R3, R5, R6, R7, R8, R9.

[0122] In other examples, a rule can be said to be efficient when it achieves a certain compression score for the given batch L. Such a score can be calculated by multiplying the compression performed by a rule, as shown in [Fig. 3], by the number of packets in the given batch L for which the rule achieves the best compression. The score can also be calculated as the score achieved by the compression rule for any packet shown in [Fig. 3], without taking into account the packets in batch L for which the rule achieves the best compression.

[0123] In some embodiments, the size S of a set of compression rules is the number of rules belonging to the set. In other examples, the size S of a set of compression rules is measured by taking into account the number and complexity of the compression operations belonging to a given rule.

[0124] In our example, a compression rule set size has been provided, which is the target size S. The goal of the process is then to run the process until a reduced rule set RRS of a size smaller than the target size is obtained. In other examples, the process simply reduces the size of the compression rule set without forcing a target size S. Suppose that the target size S, in our example, is 4 and corresponds to the number of rules in a compression rule set. The reduced rule set RRS then has a size of 7, and the process will continue in our example. The target size S will often be equal to a power of 2 in order to encode the identifier of a rule belonging to a compression rule set using a minimum number of bits. Here, with a target size S of 4, two bits are sufficient to encode the compression rule identifiers.

[0125] To this end, in our example, a compression penalty is determined, corresponding to the removal of at least one rule from the reduced set RRS. In our example, the arrangement of the compression rules of the RS set in a tree data structure is used. Indeed, this arrangement makes it possible to efficiently find rules that are more general than a given rule and that could therefore replace it to compress packets of the given batch L if a given rule were removed.

[0126] For example, if rule R6 were removed, the process can read from the data structure shown in [Fig. 3] that this rule R6 is the most efficient for packets pl and p4. If rule R6 were removed, the parent rule R3 in the tree would become the most efficient rule for these two packets pl and p4. The compression penalty attached to the removal of rule R6 for the given batch L would then include the compression of the two packets pl and p4 by rule R3 instead of rule R4. When rule R3 compresses a packet, it erases 16 bits instead of the 21 bits erased by rule R4. The compression penalty for removing rule R4 can be obtained in our example as 10 bits (twice the five fewer bits compressed on packets pl and p4 when replacing R4 with R3).

[0127] Finally, in this example, the lowest compression penalty is obtained by removing rule R9, which is then replaced by rule R8 as the rule providing the best compression for packet p8. The calculated compression penalty is only 1 bit. Note that rule R8 is present in the reduced RRS set.

[0128] Other compression penalty calculations can be carried out, not taking into account the number of packets in the given batch L for which a given rule achieves the best compression but just the compression score attached to each rule.

[0129] If the target size S is 4, the removal of rule R9 from the reduced set RRS brings its size, measured in number of rules, to 6 and two additional removals must be found.

[0130] In the example of [Fig.3], the lowest compression penalty according to the The previous calculation will be found, at this stage of the process execution, by removing rules R7 and R8 from the reduced RRS set and adding rule R4, which had been previously removed because it was not among the effective rules. The compression penalty for this operation is, firstly, 4 bits corresponding to the replacement of rule R7 by rule R4, which reduces the compression performed on packets p2 and p3 by 2 bits, and secondly, 1 bit corresponding to the replacement of rule R8 by rule R4, which reduces the compression of packet p7 by 1 bit. The total compression penalty is therefore 5 bits, and this is the lowest penalty that can be found in this example at this stage of the process.

[0131] The reduced RRS set after the removal of rules R7 and R8 and the reintroduction of rule R4 will then consist of rules RI, R3, R4, R5, and R6, each with a size of 5, which is greater than the target size S, which is 4 in our example. The final removal corresponding to the minimum compression penalty will then be to replace rule R6 with rule R3, resulting in a penalty of 10 bits, corresponding to two times 5 bits (the difference in score between the two rules R6 and R3), the factor of 2 being due to the two packets pl and p4.

[0132] The target size S of 4 is finally achieved with a reduced set RRS which includes the compression rules RI, R3, R4, R5.

[0133] This embodiment example has shown how organizing the set of compression rules RS into a tree representing which rules are more general than others simplifies the calculation of compression penalties. Such an organization can be obtained as a side effect of machine learning the set of compression rules RS using the given set L as the training set.

[0134] Finally, it should be noted here that, in this text, the term "module" can refer to a software component, a hardware component, or a set of hardware and software components. A software component itself corresponds to one or more computer programs or subprograms, or more generally to any element of a program capable of implementing a function or set of functions as described for the modules concerned. Similarly, a hardware component corresponds to any element of a hardware assembly capable of implementing a function or set of functions for the module concerned (integrated circuit, smart card, memory card, etc.).

Claims

Demands

1. A method for managing a set (RS) of compression rules applying to data packets, called the initial set, characterized in that it comprises the following steps: A step of determining (SI) a set of compression rules, called effective rules, for a batch (L) of data packets, called the given batch; A step of obtaining (S2) a reduced set (RRS) of compression rules by removing from the initial set (RS) compression rules that do not belong to the effective set of rules for the given batch (L).

2. Management method according to claim 1 characterized in that a compression rule is determined at step (SI) as belonging to the set of effective rules when there is a package in the given lot (L) for which said compression rule achieves the greatest package compression among the compression rules of the initial set (RS).

3. Management method according to any one of claims 1 or 2, characterized in that the method further comprises the following steps: A step of obtaining (S4) a compression penalty caused by the deletion of a rule from the reduced set (RRS); A step of deleting (S5) a rule from the reduced set (RRS) which minimizes the penalty obtained (S4).

4. Management method according to claim 3 characterized in that a size (S) of a set of compression rules, referred to as the target size, is provided and in that the method includes an iteration of the steps of obtaining compression penalty (S4) and deleting (S5) a rule from the reduced set (RRS) until the size of the reduced set (RRS) is less than the target size (S).

5. Management method according to any one of claims 3 or 4, characterized in that the step of obtaining a compression penalty (S4) takes into account that a compression rule removed from the reduced set (RRS) will be replaced by another rule from the reduced set (RRS) to perform data packet compressions.

6. A management method according to claim 5 characterized in that the step of obtaining (S4) a compression penalty takes in taking into account that several compression rules can be removed from the reduced set (RRS) and replaced by another rule, present in the initial set (RS) and absent from the reduced set (RRS), and that the removal step (S5) then includes the removal of said several rules and the addition of said other rule if this removal combined with this addition minimizes the penalty obtained during the acquisition step (S4).

7. Management method according to any one of claims 3 to 6 characterized in that obtaining a compression penalty (S4) takes into account the number of packets in the given batch (L) for which the deleted rule(s) achieve the greatest compression among the rules in the reduced set (RRS).

8. Management method according to any one of claims 1 to 7 characterized in that the determination step (SI) includes the calculation for a rule of the initial set (RS) of a score (SCO) obtained by multiplying the compression achieved by said rule by the number of packets in the given lot (L) for which said rule achieves the greatest compression among the rules of the initial set (RS).

9. Management method according to claim 8 characterized in that a size (S) of a set of compression rules, referred to as the target size, is provided and in that the step of obtaining (S2) the reduced set (RRS) includes the removal of the initial set (RS) of compression rules with the lowest score (SCO) until the size of the reduced set (RRS) is less than the target size (S).

10. Management method according to any one of claims 1 to 9 characterized in that the initial set (RS) is obtained by a machine learning algorithm and that the given batch (L) was included in the training set used to obtain the initial set (RS).

11. Management entity (100) managing a set (RS) of compression rules applicable to data packets, called the initial set, comprising the following modules: Module (101) for determining a set of compression rules, called the effective rules, for a batch (L) of data packets, called the given batch; Module (102) for obtaining a reduced set (RRS) of compression rules by removing rules from the initial set (RS)

12.

13.

14.

15. compression that does not belong to the set of effective rules for the given batch (L). Management entity (100) according to claim 11 further comprising the following modules: Module (104) for obtaining a compression penalty caused by the deletion of a rule from the reduced set (RRS); Module (105) for removing a rule from the reduced set (RRS) that minimizes the penalty obtained. Telecommunications equipment comprising a management entity (100) according to one of claims 11 or 12. Computer program capable of being implemented by a management entity (100) according to one of claims 11 or 12, the program comprising code instructions which, when executed by a processor, carries out the steps of the management process defined in claim 1. Data carrier on which is stored a computer program according to claim 14 comprising a sequence of instructions for implementing the management method according to claim 1 when loaded into and executed by a processor.

Images