A confusion enhancement cell based on multi-threshold MOS transistors
By using a multi-threshold MOSFET-based confusion enhancement cell, and utilizing TVD gate circuitry and a four-input competition structure, the problem of excessive MOSFET count in multi-input confusion gates is solved, achieving high security and physical indistinguishability, making it suitable for physical layer protection in security chips.
Patent Information
- Application Number
- CN202611383614.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-08
- Publication Date
- 2026-11-10
AI Technical Summary
When constructing multi-input confusion gates, the number of MOSFETs in existing TVD gates increases exponentially, resulting in an excessive increase in area. How to further optimize confusion and control the growth of the number of MOSFETs remains an urgent problem to be solved.
The system employs a mixed-enhancement cell based on multi-threshold MOS transistors, including TVD gate circuits and a four-input competition structure. The logic function is implemented through a differential pull-down network and a subsequent cross-coupled inverter. Signal processing is performed using NMOS transistors configured with different threshold voltages, maintaining layout consistency.
Without significantly increasing the number of MOSFETs, it significantly improves the difficulty and security of reverse engineering, realizes the function of a four-input obfuscation gate, and the consistent layout structure is difficult for attackers to identify.
Smart Images

Figure CN123069462A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of integrated circuit hardware security protection, and specifically to a confusion enhancement cell based on a multi-threshold MOS transistor. Background Technology
[0002] As wafer foundry services continue to globalize, the security of integrated circuits (ICs) is becoming increasingly vulnerable. Reverse engineering analyzes the interconnections and functions between transistors to recover the gate-level netlist and some logic functions. Reverse engineering can employ techniques such as layer removal and optical reconstruction, using chemical methods or specialized equipment to decompose the integrated circuit layer by layer, or by acquiring images of each layer and using image processing techniques to reconstruct the IC's gate-level netlist from these images.
[0003] Traditional hardware camouflage techniques commonly include virtual via camouflage and doping camouflage. Virtual via technology inserts non-conductive fake vias into the chip's metal layers, making them difficult to distinguish from real vias under optical or electron microscopes, thus interfering with attackers' reverse engineering of the chip's interconnect structure. However, this method requires high precision in the location, size, and fabrication process of the fake vias, making manufacturing difficult. Furthermore, fake vias may introduce reliability risks and even cause short circuits, thus limiting their practical application.
[0004] Doping camouflage techniques, while maintaining the standard cell layout geometry, alter the threshold voltage or switching state of transistors by changing the type or concentration of dopant ions implanted. However, with the development of advanced detection methods such as scanning capacitance microscopy, attackers can now perform high-precision measurements of the doping distribution within devices. For obfuscated cells with significant doping differences, their hidden features may be identified and cracked, thus weakening the security effectiveness of such camouflage techniques in practical applications.
[0005] In recent years, with the deepening research into hardware camouflage techniques, the TVD (Threshold Voltage Defined) obfuscation gate proposed by Burak Erbagci et al. has gradually become a representative new obfuscation structure. TVD gates introduce MOSFETs with different threshold voltages into the same layout structure, making different logic functions appear consistent in physical layout, thereby weakening the attacker's ability to perform reverse engineering based on layout structure and interconnection relationships. Compared with general obfuscation methods, TVD gates mainly utilize the differences in electrical behavior caused by the threshold voltage of MOSFETs to define logic functions, making it difficult for attackers to determine their true function solely through conventional physical observation. This method has smaller doping variations, is more difficult to distinguish, significantly increases the difficulty of reverse engineering, and is also more convenient in terms of process implementation.
[0006] While TVD gates excel in improving hardware security, current architectures result in an exponential increase in the number of MOSFETs and excessive area when constructing multi-input obfuscating gates to enhance obfuscation. How to further optimize the obfuscation of TVD gates while controlling the increase in the number of MOSFETs remains a pressing issue. Summary of the Invention
[0007] To address the technical problem of how to further optimize the confusion of TVD gates and control the increase in the number of MOSFETs in existing technologies, this invention proposes a confusion-enhanced cell based on multi-threshold MOSFETs, the technical solution of which is as follows:
[0008] A confusion enhancement cell based on a multi-threshold MOS transistor includes a TVD gate circuit and a four-input competition structure;
[0009] The TVD gate circuit includes a differential pull-down network and a subsequent cross-coupled inverter. The differential pull-down network is used to cover the discharge path under different input combinations and includes multiple discharge branches. Each discharge branch consists of two NMOS transistors connected in series. The subsequent cross-coupled inverter is used to amplify the voltage difference of the differential nodes and form a stable latched output.
[0010] The four-input competition structure is used to select two valid signals from four signals as the input signals of the TVD gate circuit; the four-input competition structure forms a competition node by shorting the output terminals of inverters with different threshold voltage configurations, and determines the output signal based on the competition result of the driving strength.
[0011] Furthermore, the differential pull-down network has a symmetrical structure with four discharge branches on each of the left and right sides.
[0012] Furthermore, the NMOS transistors connected in series in each discharge branch are configured as low threshold voltage devices or high threshold voltage devices; the TVD gate circuits achieve different logic function mappings through different threshold voltage configurations, and the layout structure and device connection relationship remain consistent.
[0013] Furthermore, when the TVD gate circuit implements the XOR2 function, the four discharge branches on the left and right sides of the differential pull-down network correspond to the input combinations 00, 01, 10 and 11, respectively.
[0014] Furthermore, in the discharge branch on the right, the NMOS transistors corresponding to input combinations 00 and 11 are set to high threshold voltage, and the NMOS transistors corresponding to input combinations 01 and 10 are set to low threshold voltage. The threshold voltage configuration of the discharge branch on the left is the opposite of that on the right.
[0015] Furthermore, the operation of the TVD gate circuit includes a pre-charging phase and an evaluation phase;
[0016] During the pre-charge phase, when the CLK pulse is low, the PMOS pre-charge transistor is turned on, the differential node is charged to VDD, and the node reset is completed.
[0017] During the evaluation phase, when the CLK pulse goes high, the PMOS precharge transistor turns off, and the input signal drives the differential pull-down network into a competitive discharge state. Each discharge branch generates a different equivalent current due to different threshold voltage configurations, forming an asymmetric current difference. This current difference causes the differential node voltage to change and drives the subsequent cross-coupled inverter to flip rapidly, latching to obtain a stable output.
[0018] Furthermore, the four-input competition structure includes a two-stage competition structure. The first-stage competition structure processes the signals A_1, A_2, B_1, and B_2 through inverters configured with different threshold voltages four times to obtain intermediate signals A1, A2, B1, and B2. The second-stage competition structure processes the intermediate signals A1, A2, B1, and B2 through inverters configured with different threshold voltages twice to obtain the input signals IN_1 and IN_2.
[0019] Furthermore, the outputs of the various competitive structures are shaped into standard digital levels by subsequent inverters.
[0020] Furthermore, the four-input competition structure achieves different signal selections by changing the threshold voltage configuration of the inverter, and the layout structure and device connection relationship remain consistent.
[0021] Beneficial effects
[0022] 1. In terms of physical indistinguishability, the different two-input logic gates implemented by this invention maintain completely consistent metal layer connections, and there are no differences except for the difference in threshold voltage of some MOS transistors. It is difficult for attackers to determine the threshold configuration and true function based solely on ordinary physical imaging. The implementation of reverse engineering is very difficult and is expected to increase the time and technical cost of reverse engineering.
[0023] 2. In terms of area friendliness, the mixing unit of the present invention can change the original two-input mixing gate into a four-input mixing gate by increasing the number of MOS transistors by less than 1.1 times. If the original TVD gate is expanded in the same way, although it can achieve more functions after being expanded into a four-input mixing gate, the number of MOS transistors required increases by more than 3.5 times.
[0024] 3. In terms of security, although the present invention can still achieve a two-input function, it further obfuscates the inputs on the basis of the original, thereby improving the security of the obfuscation gate. Attached Figure Description
[0025] Figure 1 This is a schematic diagram of a scrambling enhancement cell based on a multi-threshold MOS transistor, using XOR2 as an example;
[0026] Figure 2 This is a schematic diagram of the simulation results of a scrambled enhancement cell based on a multi-threshold MOS transistor, using XOR2 as an example.
[0027] Figure 3a This is a schematic diagram of the first-level competition structure of the four-input competition structure in this embodiment;
[0028] Figure 3b This is a schematic diagram of the second-level competition structure of the four-input competition structure in this embodiment;
[0029] Figure 4 This is a schematic diagram of the simulation results of the four-input competition structure in this embodiment. Detailed Implementation
[0030] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.
[0031] like Figure 1 As shown, the scrambling enhancement cell based on a multi-threshold MOS transistor, taking XOR2 as an example in this embodiment, includes a TVD gate circuit and a four-input competition structure.
[0032] The TVD gate circuit includes a symmetrical differential pull-down network with four discharge branches on each side. Each discharge branch consists of two NMOS transistors connected in series, corresponding to the input combinations 00, 01, 10, and 11 of the input signals IN_1 and IN_2, respectively. IN_1 and IN_2 are inverted by inverters to obtain IN_1_INV and IN_2_INV. In the right differential pull-down network, the NMOS transistors connected to IN_1 / IN_2 and IN_1_INV / IN_2_INV are configured with high threshold voltage (HVT), while the NMOS transistors connected to IN_1 / IN_2_INV and IN_1_INV / IN_2 are configured with low threshold voltage (LVT). The left differential pull-down network uses the completely opposite threshold voltage configuration to the right.
[0033] The TVD gate circuit includes a post-stage cross-coupled inverter, which amplifies the voltage difference between the differential nodes INT and INT_INV of the TVD gate circuit and forms a stable latched output, thereby realizing the logic function of XOR2.
[0034] The operating cycle of the TVD gate circuit is controlled by the input CLK pulse signal: when CLK=0, it is the charging period, during which the PMOS precharge transistor is turned on, and the differential nodes INT and INT_INV are quickly charged to the VDD level, thereby completing the node reset and preparing for subsequent evaluation; when CLK=1, it is the evaluation period, the PMOS precharge transistor is turned off, and the input signal drives the differential pull-down network into a competitive discharge state. At this time, the NMOS transistors of the differential pull-down network are turned on. Since the threshold voltage configurations of the NMOS transistors in different discharge branches are different, different equivalent currents are generated, forming an asymmetric current difference. The asymmetric current difference This further causes voltage changes in the differential nodes INT and INT_INV, and drives the subsequent cross-coupled inverter to flip rapidly, eventually latching to the expected level.
[0035] like Figure 2 As shown, the simulation results of the aliasing enhancement cell in this embodiment, taking XOR2 as an example, are as follows: For the timing arc h->i, after CLK goes high, the evaluation period begins. At this time, IN_1 is 0 and IN_2 is 1. Theoretically, since the NMOS transistor corresponding to the input combination 01 in the right differential pull-down network is low threshold (LVT), a strong conduction path is formed, and the differential node INT is quickly pulled down to low level. The left differential pull-down network is weakly conducted due to the high threshold (HVT) NMOS transistor, and the differential node INT_INV remains high. Finally, the output OUT jumps from 0 to 1, and OUT_INV jumps from 1 to 0, which is consistent with the expected result and realizes the XOR2 function.
[0036] like Figures 3a to 3b As shown, the four-input competition structure in this embodiment is used to select two valid signals as the input signals IN_1 and IN_2 from four signals A_1, A_2, B_1, and B_2. Its core is that when the inverters corresponding to the two inputs produce the same driving direction to the common node, the common node will be stably pulled to a high or low level. When the inverters corresponding to different inputs produce opposite driving directions to the common node, current competition will occur on the common node, making it a competing node. In this case, the inverter branch with the lower threshold voltage will provide a larger pull-up or pull-down current, thus playing a dominant role and determining whether the competing node ultimately approaches a high or low level. The determination of valid inputs is achieved through competition based on drive strength rather than explicit selection of control signals.
[0037] like Figure 3aAs shown, in the structure for generating intermediate signal A1, A_1 is connected to a high-threshold inverter, A_2 is connected to a low-threshold inverter, and the effective signal is A_2; in the structure for generating intermediate signal A2, A_1 is connected to a low-threshold inverter, A_2 is connected to a high-threshold inverter, and the effective signal is A_1; in the structure for generating intermediate signal B1, B_1 is connected to a high-threshold inverter, B_2 is connected to a low-threshold inverter, and the effective signal is B_2; in the structure for generating intermediate signal B2, B_1 is connected to a low-threshold inverter, B_2 is connected to a high-threshold inverter, and the effective signal is B_1.
[0038] like Figure 3b As shown, in the structure for generating the input signal IN_1, the intermediate signal A1 is connected to a low-threshold inverter, the intermediate signal B1 is connected to a high-threshold inverter, and the effective signal is A1; in the structure for generating the input signal IN_2, the intermediate signal A2 is connected to a high-threshold inverter, the intermediate signal B2 is connected to a low-threshold inverter, and the effective signal is B2.
[0039] The outputs of each competing structure are shaped by the inverter in the next stage, restoring the non-ideal intermediate level that may exist at the competing node to the standard digital level, while improving the driving capability and noise margin of subsequent stages.
[0040] like Figure 4 As shown in the simulation results of the four-input competition structure in this embodiment, the final actual effective signals of the four-input competition structure are A_2 and B_1, which correspond to the data of the input signals IN_1 and IN_2 respectively, which is consistent with the expected results.
[0041] In summary, this invention has made progress in terms of physical obfuscation effect, area friendliness, and security, laying a solid foundation for the practical deployment of obfuscated enhancement cells based on multi-threshold voltage MOS transistors in large-scale chips.
[0042] The aliasing enhancement cell based on multi-threshold MOS transistors of this invention is suitable for physical layer protection of security chips (such as encryption chips and payment chips) and intellectual property protection. This invention is implemented based on the 55nm CMOS (Complementary Metal-Oxide-Semiconductor) process node and can be widely used in consumer electronics, aerospace and military industries, Internet of Things security and other fields.
[0043] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A confusion enhancement cell based on a multi-threshold MOS transistor, characterized in that: Including TVD gate circuits and four-input competitive structures; The TVD gate circuit includes a differential pull-down network and a subsequent cross-coupled inverter. The differential pull-down network is used to cover the discharge path under different input combinations and includes multiple discharge branches. Each discharge branch consists of two NMOS transistors connected in series. The subsequent cross-coupled inverter is used to amplify the voltage difference of the differential nodes and form a stable latched output. The four-input competition structure is used to select two valid signals from four signals as the input signals of the TVD gate circuit; the four-input competition structure forms a competition node by shorting the output terminals of inverters with different threshold voltage configurations, and determines the output signal based on the competition result of the driving strength.
2. The aliasing enhancement cell based on a multi-threshold MOS transistor as described in claim 1, characterized in that: The differential pull-down network has a symmetrical structure with four discharge branches on each of the left and right sides.
3. The confusion enhancement cell based on a multi-threshold MOS transistor as described in claim 2, characterized in that: The NMOS transistors connected in series in each discharge branch are configured as low threshold voltage devices or high threshold voltage devices; the TVD gate circuits achieve different logic function mappings through different threshold voltage configurations, and the layout structure and device connection relationship remain consistent.
4. The confusion enhancement cell based on a multi-threshold MOS transistor as described in claim 3, characterized in that: When the TVD gate circuit implements the XOR2 function, the four discharge branches on the left and right sides of the differential pull-down network correspond to the input combinations 00, 01, 10 and 11, respectively.
5. The aliasing enhancement cell based on a multi-threshold MOS transistor as described in claim 4, characterized in that: In the discharge branch on the right, the NMOS transistors corresponding to input combinations 00 and 11 are set to high threshold voltage, and the NMOS transistors corresponding to input combinations 01 and 10 are set to low threshold voltage. The threshold voltage configuration of the discharge branch on the left is the opposite of that on the right.
6. The aliasing enhancement cell based on a multi-threshold MOS transistor as described in claim 1, characterized in that: The working process of the TVD gate circuit includes a pre-charging stage and an evaluation stage; During the pre-charge phase, when the CLK pulse is low, the PMOS pre-charge transistor is turned on, the differential node is charged to VDD, and the node reset is completed. During the evaluation phase, when the CLK pulse goes high, the PMOS precharge transistor turns off, and the input signal drives the differential pull-down network into a competitive discharge state. Each discharge branch generates a different equivalent current due to different threshold voltage configurations, forming an asymmetric current difference. This current difference causes the differential node voltage to change and drives the subsequent cross-coupled inverter to flip rapidly, latching to obtain a stable output.
7. The scrambling enhancement cell based on a multi-threshold MOS transistor as described in claim 1, characterized in that: The four-input competition structure includes a two-stage competition structure. The first-stage competition structure processes the signals A_1, A_2, B_1, and B_2 through inverters configured with different threshold voltages four times to obtain intermediate signals A1, A2, B1, and B2. The second-stage competition structure processes the intermediate signals A1, A2, B1, and B2 through inverters configured with different threshold voltages twice to obtain the input signals IN_1 and IN_2.
8. The scrambling enhancement cell based on a multi-threshold MOS transistor as described in claim 7, characterized in that: The outputs of each stage of the competitive structure are shaped into standard digital levels by the subsequent inverter.
9. The aliasing enhancement cell based on a multi-threshold MOS transistor as described in claim 1, characterized in that: The four-input competition structure achieves different signal selections by changing the threshold voltage configuration of the inverter, and the layout structure and device connection relationship remain consistent.
