The Research For Semi-invasive Extracting Information From Non-volatile Memories

With the development of integrated circuit technology,all kinds of portable electronic devices can be found everywhere in life,such as smart phone,smartcards,notebook computer,etc.These electronic products bring a variety of convenience to people's life,while the information security of equipments is increasingly important.Hardware is the carrier for storing,transmitting and processing information,the core issue of hardware security is the chip storage security.Semiconductor chips generally contain one or more non-volatile memories,such as EEPROM and Flash.Non-volatile memory,as a storage medium,is normally featured with high security,and thus often contains sensitive data such as personal information,passwords,cryptographic keys and other confidential contents.Therefore,the assurance of the confidentiality and integrity of sensitive information of non-volatile memory is an important measurement and basic requirement of semiconductor chips security.In this paper,the semi-invasive attack method was used to extract information from non-volatile memory.Optical fault injection would be executed to extract sensitive data from the flash memory of the smartcard,and detect the ability against semi-invasive attacks of non-volatile memory.Optical fault injection attack is a conventional way that aims at extracting the information stored on memories.However,it is difficult to use this method to attack flash memories of modern micro smartcards as several reasons such as modern deep submicron technologies,smaller feature size and multiple metal layers.In this paper,the attack on flashes with optical fault injection is analyzed theoretically,? he actual attack method is proposed,and the experimental process is presented in detail.In the experiment,the target chip is a smartcard chip of domestic production,with unpack process executed,verify operation should be added to the smartcard based on its own performance function.RISCUE device is used to complete automated optical fault injection when the chip performs verify operation.Experimental results show that some certain bytes of data can be changed into known value with optical fault injection attacks.In the read/write process,some secure leakage information can be obtained by this approach of attack.Furthermore,optical fault injection may affect the verify operation in certain points sensitive to laser or directly impact the returning flag bit to pass the verify operation.The feature of the fault injection attack on flashes shown in the experiment can be utilized to bypass the verify operation of certain changed bytes and to guess the remaining original bytes to gain contents stored in non-volatile memories.Overall,using semi-invasive attack methods to extract contents from non-volatile memories is difficult as its continuous engineering problems,but the further research can be attached to this method because they are direct and effective,and complement the non-invasive attack methods.Meanwhile,studies have shown that non-volatile memories have certain ability to resist semi-invasive attacks.