Most administrators are concerned with the largest possible keys that will guarantee security. But if they need to deal with the relationship between key sizes (key material), the comparison is difficult for them. What is better? How to compare a 128b key for symmetric ciphers with 4096b RSA? What should be more secure? Or the current question, how to compare the strength of classical and quantum-resistant algorithms? What affects the evaluation?
Until the advent of quantum computers, mathematics clearly dictated the minimum key sizes to us in roughly the following way. Based on the properties of the algorithm or the best known attack, the total complexity, expressed as the number of operations, was calculated. Since this complexity is expressed as a power of two, the exponent is an expression of the security equivalent and security.
If this output is compared with the official recommendations, there are some minor differences. These are given by the algorithm architecture, implementation, specific properties of the algorithm and, if applicable, a safety margin. These are precisely the reasons that lead to an approximately 20% - 30% increase in the required key size for asymmetric algorithms, as stated in the documentation of BSI, ECRYPT, NIST, NUKIB and others. And this is also the reason why it is advisable to follow such advice. Relying solely on complexity can be shortsighted, and especially for asymmetric algorithms, it may not fully reflect reality.
| Security equivalent | RSA (bits) | DLP-MG (bits) | DLP-AG (bits) | Hash (bits) | Symmetric algorithms (bits) |
| 64 | 355 | 600 | 129 | 128 | 64 |
| 92 | 812 | 1392 | 185 | 184 | 92 |
| 112 | 1284 | 2218 | 225 | 224 | 112 |
| 128 | 1758 | 3052 | 257 | 256 | 128 |
| 160 | 2993 | 5234 | 321 | 320 | 160 |
| 192 | 4648 | 8173 | 385 | 384 | 192 |
| 224 | 6766 | 11949 | 449 | 448 | 224 |
| 256 | 9389 | 16642 | 513 | 512 | 256 |
| 384 | 25715 | 46029 | 769 | 768 | 384 |
| 512 | 53100 | 95636 | 1025 | 1024 | 512 |
The above table shows at first glance how the RSA algorithm, based on factorization, performed. In the case of the DLP-MG column (discrete logarithm problem over multiplicative group), this covers the Diffie-Hellman algorithm, but also applies to the ElGamal algorithm, the Schnorr signature, and the DSA algorithm. The next column is DLP-AG (discrete logarithm problem over additive group). This includes algorithms such as ECDH, ECDSA, and EDDSA. Next are hashes sorted by output width, and finally symmetric algorithms, under which block and stream algorithms such as AES, CHACHA20, and others can be classified. For symmetric algorithms, only the influence of the key length is considered.
The arrival of quantum computers changes everything. The original idea of security has taken its toll, most algorithms have been taken by the ears in terms of security, as they say. Current asymmetric algorithms should be broken by Shor's algorithm, while Grover's algorithm is a threat to symmetric algorithms with small key widths. For this reason, it makes no sense to discuss the asymmetric algorithms in the previous table (RSA, Diffie-Hellman, ElGamal, Schnorr signature, DSA, ECDH, ECDSA and EDDSA), because the complexity of the attack is extremely low. Such an attack would take place in the case of a cryptographically relevant quantum computer in the order of minutes to hours. On the other hand, new algorithms are emerging that are ready to withstand these attacks. The first overview therefore describes how the security of hash functions with a certain output width and similarly symmetric algorithms with a specific key length will change.
Key width as such may not be sufficient protection in the future. The current situation somewhat reminds me of the situation with the DES and 3DES algorithms. In the years 1995-2000, due to the increasing data volumes, the risks of algorithms with a block size of 64 bits were considered. In order to protect against some attacks, a double or triple key width was used as a temporary solution. Therefore, two variants of the 3DES algorithm were created, where the first performed encryption using three separate keys (Enc k1, Enc k2, Enc k3), the second used only two keys and worked in EDE mode (Enc k1, Dec k2, Enc k1). Fortunately, Groover's attack cannot currently be used in this way. Unfortunately, there is a similar algorithm called Brassard–Høyer–Tapp. It is designed to solve this problem. Fortunately, its demand for quantum memory is so high that it will not be possible to satisfy it in the coming decades. Unlike this algorithm, Groover's algorithm practically does not require any memory. However, if in the future an attack is found where the BHT algorithm does not need memory, we will have serious problems.
| Security equivalent | Hash functions (bits) | Symmetric algorithms (bits) |
| 32 | 128 | 64 |
| 64 | 184 | 128 |
| 92 | 224 | 184 |
| 128 | 256 | 256 |
| 160 | 320 | 320 |
| 192 | 384 | 384 |
| 224 | 448 | 448 |
| 256 | 512 | 512 |
| 384 | 768 | 768 |
| 512 | 1024 | 1024 |
The table shows the change in the difficulty of attacking quantum computers compared to digital computers (previous tables). In order to ensure adequate security, it is necessary to double the width of the hash function output, and the key material entering symmetric algorithms must also be twice as large.
Upcoming algorithms resistant to attacks using quantum computers (PQC/QRC) can help solve the problem with asymmetric algorithms. These are mainly sets standardized by IETF and NIST. The following note needs to be added to this. Sometimes you can come across the term quantum algorithms, which refers to attacks using Groover's and Shor's algorithms. These are not algorithms designed for encryption on quantum computers. Similarly, quantum cryptography is not a matter of algorithms for encryption on quantum computers, but of securing transmissions using quantum phenomena. These, under specific conditions, protect the communication channel from eavesdropping or message modification.
New standardized algorithms are based on various approaches. On the one hand, there are algorithms built on lattices, linear codes, systems of quadratic equations with multiple unknowns, supersingular curves and hash trees. There are other interesting problems, but for them either practically there are no algorithms,because they were found to be insufficient. At this point, we have, thanks to NIST, standardized algorithms ML-KEM, ML-DSA, FN-DSA, and FN-DSA are being prepared for standardization. The IETF proposes XMSS, XMSS-MT, and LMS. There are a large number of comparisons of these asymmetric algorithms with classical ones, and often the comparisons are completely meaningless. This criterion must not be guided by the key width, but only by the security equivalent. Using key width is a misunderstanding of the differences between algorithms and the causes that lie in the problem of complexity.
| Security equivalent | Security level | Description |
| 128 bits | Security level 1 | AES-128, exhaustive key search |
| 128 bits | Security level 2 | SHA-256, collision search |
| 192 bits | Security level 3 | AES-192, exhaustive key search |
| 192 bits | Security level 4 | SHA-384, collision search |
| 256 bits | Security level 5 | AES-256, exhaustive key search |
According to the above overview, it is possible to translate the corresponding security equivalent for example, to the RSA algorithm. This means the need to use the corresponding complexity and resistance of the algorithms, not the length of the private key. The meaning of such a comparison would be meaningless.
As part of the standardization of PQC algorithms, the FIPS-203 ML-KEM standard (originally CRYSTALS Kyber) was created. Also currently in the process of standardization is FIPS-207 HQC-KEM (HQC algorithm).
| Algorithm | Security level | Private key (bits) | Public key (bits) |
| ML-KEM-512 | 1 | 6400 | 13056 |
| ML-KEM-768 | 2 | 9472 | 19200 |
| ML-KEM-1024 | 3 | 12544 | 25344 |
| HQC-128 | 1 | 448 | 19272 |
| HQC-192 | 2 | 512 | 36808 |
| HQC-256 | 3 | 576 | 57960 |
As part of the standardization of PQC algorithms, the FIPS-204 ML-DSA (originally CRYSTALS Dilithium) and FIPS-205 SLH-DSA (SPHINCS+) standards have been created. Furthermore, the FIPS-206 FN-DSA (originally Falcon) is currently in the process of being standardized. Other algorithms for the second round are in the selection phase.
| Algorithm | Security level | Private key (bits) | Public key (bits) |
| ML-DSA-44 | 1 | 20480 | 10496 |
| ML-DSA-65 | 2 | 22256 | 15616 |
| ML-DSA-87 | 3 | 39168 | 20736 |
| SLH-DSA-128 | 1 | 512 | 256 |
| SLH-DSA-192 | 2 | 512 | 256 |
| SLH-DSA-256 | 3 | 512 | 256 |
| FN-DSA-512 | 1 | 10248 | 7176 |
| FN-DSA-768 | 2 | 14856 | 10760 |
| FN-DSA-1024 | 3 | 18440 | 14344 |
In addition to the algorithms standardized by NIST, algorithms for digital signatures have also been created and standardized within the IETF. This table is not a complete list of all their variants.
| Algorithm | Security level | Private key | Public key (bits) |
| LMS-128 | 1 | By tree size | 256 |
| XMSS-128 | 1 | By tree size | 256 |
| XMSS-192 | 2 | By tree size | 256 |
| XMSS-256 | 3 | By tree size | 256 |
| XMSSMT-128 | 1 | By tree size | 256 |
| XMSSMT-192 | 2 | By tree size | 256 |
| XMSSMT-256 | 3 | By tree size | 256 |
The same is true with the implementation of technologies using cryptographic algorithms. The level of resistance of all algorithms protecting data should be similar. Significant deviations can cause unnecessarily high overhead in some areas (key agreement, integrity, confidentiality, authentication). Unnecessary overuse of resources generates heat, at which point overpriced direct heaters are created instead of data protection. It is therefore appropriate for all components to use approximately the same security equivalent, and its strengthening in certain areas should be supported by a developed risk analysis.
Comparing the incomparable is the most common problem of changes brought about by the transition to quantum resistant cryptography. The imbalance of algorithms in terms of security equivalent or insufficient strength of protection using encryption technologies has a similar effect. The above approaches can compromise data, especially ensuring its confidentiality and integrity.
1. Introductory Provisions
1.1. These General Terms and Conditions are, unless otherwise agreed in writing in the contract, an integral part of all contracts relating to training organised or provided by the trainer, Jan Dušátko, IČ 434 797 66, DIČ 7208253041, with location Pod Harfou 938/58, Praha 9 (next as a „lector“).2. Creation of a contract by signing up for a course
2.1. Application means unilateral action of the client addressed to the trainer through a data box with identification euxesuf, e-mailu with address register@cryptosession.cz or register@cryptosession.info, internet pages cryptosession.cz, cryptosession.info or contact phone +420 602 427 840.3. Termination of the contract by cancellation of the application
3.1. The application may be cancelled by the ordering party via e-mail or via a data mailbox.4. Price and payment terms
4.1. By sending the application, the ordering party accepts the contract price (hereinafter referred to as the participation fee) indicated for the course.5. Training conditions
5.1. The trainer is obliged to inform the client 14 days in advance of the location and time of the training, including the start and end dates of the daily programme.6. Complaints
6.1. If the participant is grossly dissatisfied with the course, the trainer is informed of this information.7. Copyright of the provided materials
7.1. The training materials provided by the trainer in the course of the training meet the characteristics of a copyrighted work in accordance with Czech Act No 121/2000 Coll.8. Liability
8.1. The trainer does not assume responsibility for any shortcomings in the services of any third party that he uses in the training.9. Validity of the Terms
9.1 These General Terms and Conditions shall be valid and effective from 1 October 2024.Consent to the collection and processing of personal data
According to Regulation (EU) No 2016/679 of the European Parliament and of the Council on the protection of individuals with regard to the processing of personal data and on the free movement of such data and repealing Directive 95/46/EC (General Data Protection Regulation, hereinafter referred to as "the Regulation"), the processor xxx (hereinafter referred to as "the Controller") processes personal data. Individual personal data that are part of the processing during specific activities at this web presentation and in the course of trade are also broken down.Information about the records of access to the web presentation
This website does not collect any cookies. The site does not use any analytical scripts of third parties (social networks, cloud providers). For these reasons, an option is also offered for displaying the map in the form of a link, where the primary source is OpenStreet and alternatives then the frequently used Maps of Seznam, a.s., or Google Maps of Google LLC Inc. The use of any of these sources is entirely at the discretion of the users of this site. The administrator is not responsible for the collection of data carried out by these companies, does not provide them with data about users and does not cooperate on the collection of data.Information about contacting the operator of the site
The form for contacting the operator of the site (administrator) contains the following personal data: name, surname, e-mail. These data are intended only for this communication, corresponding to the address of the user and are kept for the time necessary to fulfil the purpose, up to a maximum of one year, unless the user determines otherwise.Information about the order form
In case of an interest in the order form, the form contains more data, i.e. name, surname, e-mail and contact details for the organisation. These data are intended only for this communication, corresponding to the address of the user and are kept for one year, unless the user determines otherwise. In the event that a business relationship is concluded on the basis of this order, only the information required by Czech law on the basis of business relations (company name and address, bank account number, type of course and its price) will continue to be kept by the administrator.Information about the course completion document
Within the course, a course completion document is issued by the processor. This document contains the following data: student's name and surname, the name and date of the course completion and the employer's name. The information is subsequently used for the creation of a linear hash tree (non-modifiable record). This database contains only information about the provided names and company names, which may or may not correspond to reality and is maintained by the processor for possible re-issuance or verification of the document's issuance.Rights of the personal data subject
The customer or visitor of this website has the possibility to request information about the processing of personal data, the right to request access to personal data, or the right to request the correction or deletion of any data held about him. In the case of deletion, this requirement cannot be fulfilled only if it is not data strictly necessary in the course of business. The customer or visitor of this website also has the right to obtain explanations regarding the processing of his personal data if he finds out or believes that the processing is carried out in violation of the protection of his private and personal life or in violation of applicable legislation, and the right to request removal of the resulting situation and to ensure the correction.