3/27/23, 5:46 PM
`
`Evaluation of Power Efficiency for Digital Serial Interfaces of Microcontrollers | IEEE Conference Publication | IEEE Xplore
`
`IEEE.org
`
`IEEE Xplore
`
`IEEE SA
`
`IEEE Spectrum
`
`More Sites
`
`Cart
`
`
`Create
`Account
`
`
`Personal
`Sign In
`
`
`
`Browse  My Settings  Help 
`
`Access provided by:
`Irell & Manella LLP
`
`File Cabinet
`
`Sign Out
`Show Usage 
`
`Access provided by:
`Irell & Manella LLP
`
`File Cabinet
`
`Sign Out
`Show Usage 
`
`All
`
`
`
`
`
`ADVANCED SEARCH
`
`Conferences > 2012 5th International Confer... 
`
`Evaluation of Power Efficiency for Digital Serial Interfaces of Microcontrollers
`Publisher: IEEE
`
`Cite This
`
` PDF
`
`A
`
`lerts
`
`Manage Content Alerts
`Add to Citation Alerts
`
`  
`
`Konstantin Mikhaylov ; Jouni Tervonen All Authors 
`
`12
`Paper
`Citations
`
`841
`Full
`Text Views
`
`Abstract
`
`Document Sections
`
`I.
`
`Introduction
`
`II. Serial Interfaces
`
`III. Experiment Setup
`
`IV. Results
`
`V. Discussion and
`Conclusion
`
`Authors
`
`Figures
`
`References
`
`Citations
`
`Keywords
`
`Abstract:Over the recent years, novel low-power microcontrollers have been introduced. This has allowed the
`development of various applications, which can operate over long period... View more
`
` Metadata
`Abstract:
`Over the recent years, novel low-power microcontrollers have been introduced. This has allowed the development of
`various applications, which can operate over long periods of time and fulfill their tasks having very limited amount of
`energy, such as e.g., Wireless Sensor Networks (WSN).Nonetheless, for fulfilling their tasks such devices in addition to
`the central processor often have to include several microcontrollers or some peripherals, such as sensors, external
`memory chips or other application-specific hardware. All those peripherals, except the energy for actual operation also
`require some energy for communicating to the central processor. In the paper we are investigating and comparing the
`energy consumption for three most widely used embedded systems' digital serial interfaces, namely I2C, SPI and
`UART. The presented results have been obtained using the real-life microcontroller (PIC18F family from Microchip) and
`reveal the energy consumption for different interface implementation methods (hardware vs. software) and various
`scenarios (idle interface, different data transmit or receive scenarios). The presented data can be valuable as well for
`researchers and engineers and allow to choose the most energy efficient communication interface and its
`implementation method to be used in the most energy-critical applications.
`
`https://ieeexplore.ieee.org/document/6208716
`
`1/4
`
`
`
` D
`
`ownl
`
`PDF
`
`Netlist Ex 2052
`Samsung v Netlist
`IPR2022-00996
`
`

`

`3/27/23, 5:47 PM
`Metrics
`
`Evaluation of Power Efficiency for Digital Serial Interfaces of Microcontrollers | IEEE Conference Publication | IEEE Xplore
`Published in: 2012 5th International Conference on New Technologies, Mobility and Security (NTMS)
`
`More Like This
`
`Footnotes
`
`Date of Conference: 07-10 May 2012
`
`INSPEC Accession Number: 12770882
`
`Date Added to IEEE Xplore: 31 May 2012
`
`DOI: 10.1109/NTMS.2012.6208716
`
` ISBN Information:
`
` ISSN Information:
`
`Publisher: IEEE
`
`Conference Location: Istanbul, Turkey
`
` Contents
`
`I. Introduction
`The technology advances of the recent years allowed to develop wide range of novel embedded
`systems with extremely low power consumption. Nowadays, the low-power embedded processors
`are widely used as core processing systems in various portable consumer electronics, including
`communication devices, computers and medical devices [1], [2]. Nonetheless, although the
`embedded processors themselves often have quite high power efficiency, the peripheral
`components (e.g., sensors, external memory chips or cards, graphical user interfaces hardware)
`that are required in many applications could significantly increase the energy consumption for the
`whole device. The energy, that is consumed by those peripherals consists as well of the energy for
`their actual operation and the energy for peripheral communication with core processor. As has
`been revealed e.g., in [3], sometimes the energy consumption for the communication with a
`Continue Reading
`peripheral appears to be even higher than the one for this peripheral operation. Especial
`importance the problem of energy efficiency overall and energy efficiency for communication in
`particular achieves for applications, where continues autonomous operation is required, such as
`e.g., many real-life applications of Wireless Sensor Networks (WSN) [4] or personal communication
`devices with multiple processors [5]. Although some disparate scraps of information can be found in
`the works focusing the energy efficiency of the WSN platforms and other portable communication
`devices (see e.g., [3], [5]), the actual problem of energy efficiency for the most widespread
`embedded systems' communication interfaces, to the best of our knowledge, has not been
`considered previously at all. Therefore, in this paper, we evaluate the most widely used digital
`interfaces and compare them from the point of view of energy efficiency.
`
`Authors
`
`Figures
`
`References
`
`Citations
`
`Keywords
`
`Metrics
`
`Footnotes
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`https://ieeexplore.ieee.org/document/6208716
`
`2/4
`
`Netlist Ex 2052
`Samsung v Netlist
`IPR2022-00996
`
`

`

`3/27/23, 5:47 PM
`
`Evaluation of Power Efficiency for Digital Serial Interfaces of Microcontrollers | IEEE Conference Publication | IEEE Xplore
`
`https://ieeexplore.ieee.org/document/6208716
`
`3/4
`
`Netlist Ex 2052
`Samsung v Netlist
`IPR2022-00996
`
`More
`Like
`This
`Digital sound recorder with ARM microcontroller and SD card
`2012 20th Telecommunications Forum (TELFOR)
`Published: 2012
`Characteristics of Serial Peripheral Interfaces (SPI) Timing Parameters for Optical Mouse Sensor
`2006 IEEE International Conference on Semiconductor Electronics
`Published: 2006
`Show
`More
`

`

`3/27/23, 5:47 PM
`
`Evaluation of Power Efficiency for Digital Serial Interfaces of Microcontrollers | IEEE Conference Publication | IEEE Xplore
`
`IEEE Personal Account
`
`Purchase Details
`
`Profile Information
`
`Need Help?
`
`CHANGE
`USERNAME/PASSWORD
`
`PAYMENT OPTIONS
`
`VIEW PURCHASED
`DOCUMENTS
`
`COMMUNICATIONS
`PREFERENCES
`
`PROFESSION AND
`EDUCATION
`
`US & CANADA: +1 800 678
`4333
`
`WORLDWIDE: +1 732 981
`0060
`
`TECHNICAL INTERESTS
`
`CONTACT & SUPPORT
`
`Follow
`
`  
`
`About IEEE Xplore | Contact Us | Help | Accessibility | Terms of Use | Nondiscrimination Policy | IEEE Ethics Reporting  | Sitemap |
`IEEE Privacy Policy
`A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of
`humanity.
`
`© Copyright 2023 IEEE - All rights reserved.
`
`IEEE Account
`
`» Change Username/Password
`» Update Address
`Purchase Details
`
`» Payment Options
`» Order History
`» View Purchased Documents
`Profile Information
`
`» Communications Preferences
`» Profession and Education
`» Technical Interests
`Need Help?
`
`» US & Canada: +1 800 678 4333
`» Worldwide: +1 732 981 0060
`» Contact & Support
`
` About IEEE Xplore Contact Us
`
`|
`
`
`
`|
`
`Help
`
`
`
`|
`
`Accessibility
`
`
`
`|
`
`Terms of Use
`
`
`
`|
`
`Nondiscrimination Policy
`
`
`
`|
`
`Sitemap
`
`
`
`|
`
`Privacy & Opting Out of Cookies
`
`A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity.
`© Copyright 2023 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.
`
`https://ieeexplore.ieee.org/document/6208716
`
`4/4
`
`Netlist Ex 2052
`Samsung v Netlist
`IPR2022-00996
`
`

`

`Evaluation of Power Efficiency for Digital Serial
`Interfaces of Microcontrollers
`
`Konstantin Mikhaylov and Jouni Tervonen
`RFMedia Laboratory
`Oulu Southern Institute,University of Oulu
`Vierimaantie 5, 84100, Ylivieska, Finland
`Email: {konstantin.mikhaylov,jouni.tervonen}@oulu.fi
`
`Abstract—Over the recent years, novel low-power microcon-
`trollers have been introduced. This has allowed the development
`of various applications, which can operate over long periods of
`time and fulfill their tasks having very limited amount of energy,
`such as e.g., Wireless Sensor Networks (WSN). Nonetheless, for
`fulfilling their tasks such devices in addition to the central
`processor often have to include several microcontrollers or
`some peripherals, such as sensors, external memory chips or
`other application-specific hardware. All those peripherals, except
`the energy for actual operation also require some energy for
`communicating to the central processor. In the paper we are
`investigating and comparing the energy consumption for three
`most widely used embedded systems’ digital serial interfaces,
`namely I2C, SPI and UART. The presented results have been
`obtained using the real-life microcontroller (PIC18F family from
`Microchip) and reveal the energy consumption for different
`interface implementation methods (hardware vs. software) and
`various scenarios (idle interface, different data transmit or
`receive scenarios). The presented data can be valuable as well for
`researchers and engineers and allow to choose the most energy
`efficient communication interface and its implementation method
`to be used in the most energy-critical applications.
`Keywords-digital serial interface; energy consumption; embed-
`ded systems; power efficiency; wireless sensor networks
`
`I. INTRODUCTION
`The technology advances of the recent years allowed to
`develop wide range of novel embedded systems with extremely
`low power consumption. Nowadays, the low-power embedded
`processors are widely used as core processing systems in var-
`ious portable consumer electronics, including communication
`devices, computers and medical devices [1], [2]. Nonetheless,
`although the embedded processors themselves often have quite
`high power efficiency, the peripheral components (e.g., sen-
`sors, external memory chips or cards, graphical user interfaces
`hardware) that are required in many applications could signif-
`icantly increase the energy consumption for the whole device.
`The energy, that is consumed by those peripherals consists as
`well of the energy for their actual operation and the energy for
`peripheral communication with core processor. As has been
`revealed e.g., in [3], sometimes the energy consumption for
`the communication with a peripheral appears to be even higher
`than the one for this peripheral operation. Especial importance
`
`This paper has been accepted for publication in the proceedings of WSN-
`ADT Workshop that had been held in conjunction with NTMS 2012.
`
`the problem of energy efficiency overall and energy efficiency
`for communication in particular achieves for applications,
`where continues autonomous operation is required, such as
`e.g., many real-life applications of Wireless Sensor Networks
`(WSN) [4] or personal communication devices with multiple
`processors [5]. Although some disparate scraps of information
`can be found in the works focusing the energy efficiency of
`the WSN platforms and other portable communication devices
`(see e.g., [3], [5]), the actual problem of energy efficiency for
`the most widespread embedded systems’ communication in-
`terfaces, to the best of our knowledge, has not been considered
`previously at all. Therefore, in this paper, we evaluate the most
`widely used digital interfaces and compare them from the point
`of view of energy efficiency.
`
`II. SERIAL INTERFACES
`As revealed in [6] and [7], out of the wide range of existing
`communication interfaces, nowadays the most widely used are
`the analog interface and I2C, SPI, UART and 1-wire digital
`interfaces. Each of these interfaces has some specifics, that
`influence the data rate, energy consumption and available
`additional interface’s features (e.g., robust communication or
`device identification support). Many of those interfaces are
`nowadays already implemented by the hardware modules that
`are integrated into the majority of contemporary microcon-
`trollers [8].
`
`A. UART
`The interfaces based on Universal Asynchronous Re-
`ceivers/Transmitters (UARTs) (later in the paper those will
`be addressed as ”UART interfaces”) are often used for im-
`plementing such standards as Electronic Industries Alliance
`(EIA) Recommended Standards RS-232, RS-422 and RS-
`485. UARTs are nowadays one of the most commonly used
`method for communication between an embedded system and
`an external device [9]. As Fig. 1 revels, UARTs provide full-
`duplex asynchronous serial peer-to-peer communication. The
`data transmission over UART usually starts with a start bit(S),
`that alerts the receiver that a word of data is about to be
`sent and allows the receiver to synchronize clocks with the
`transmitter. After the start-bit, the actual data is transmitted
`starting with the least-significant bit. Once a data word has
`
`978-1-4673-0229-6/12/$31.00 ©2012 IEEE
`Authorized licensed use limited to: Irell & Manella LLP. Downloaded on March 28,2023 at 00:46:30 UTC from IEEE Xplore. Restrictions apply.
`
`Netlist Ex 2052
`Samsung v Netlist
`IPR2022-00996
`
`

`

`Fig. 1: UART interface and data format
`
`been sent, transmitter may issue the parity bit(Pr) to provide
`a simple error checking and/or a stop bit(P) to signalize the
`end of data word transmission.
`
`B. SPI
`The Serial Peripheral Interface (SPI), that has been pop-
`ularized by Motorola, is a synchronous serial interface that
`operates in full duplex mode [10]. The typical method for
`connecting several SPI slave devices to a master is presented
`in Fig. 2. As the figure shows, SPI bus utilizes three common
`lines for all slave devices: clock (SCLK); master output,
`slave input (MOSI); master input, slave output (MISO); and a
`separate chip select (CS) line for each slave device. Therefore,
`before starting the communication, the SPI master device pulls
`down CS line of the required slave device to select it. The SPI
`specification does not define neither any maximum data rate
`(for existing devices it reaches dozens MHz) nor any particular
`addressing scheme or acknowledgment mechanism.
`
`C. I2C
`The developed by Philips Inter-Integrated Circuit (I2C)
`interface and its data format are presented in Fig. 3. As this
`figure reveals, the I2C interface uses two common physical
`lines for clock (SCLK) and data (SDA), which are pulled-up
`with resistors (Rp) [11]. The I2C interface supports multiple
`slave and master devices; therefore, the master device starts
`the communication by sending the start bit(S) and the unique
`slave device address (which usually consists of 7 bits,
`although addresses of 10 bits are also defined in recent I2C
`standard revisions). Together with the slave device’s address,
`the master transmits the 1-bit Read/Write (R/W ) for defining
`the communication direction, which would be used until the
`stop bit (P) closes the current session. The I2C communication
`protocol implements per-byte acknowledgments (A/A). The
`standardized data rates for I2C devices are 10 kbit/s, 100
`kbit/s and 400 kbit/s, although most recent I2C revision [11]
`also supports the rates of 1 Mbit/s and 3.4 Mbit/s.
`
`The results of comparison for different serial interfaces are
`summarized in Table I.
`
`III. EXPERIMENT SETUP
`For evaluating the energy consumption for I2C, SPI and
`UART interfaces in real-life conditions we have used the
`testbed build around two PICKit 3 development boards [13]
`from Microchip with PIC18F45K20 microcontrollers onboard
`
`Fig. 2: SPI interface and data format
`
`Fig. 3: I2C interface and data format
`
`[14]. The microcontroller of the first board was programmed
`to transmit the data over the serial interface (TX board), while
`the microcontroller of the second board was used to receive
`the data (RX board). The pins of the microcontrollers that
`were required for implementing the serial interfaces have been
`connected together using 10 cm long wires.
`For implementing the required serial interfaces we have
`used two different options: interfaces were implemented us-
`ing the available hardware modules of the microcontroller
`(Master Synchronous Serial Port (MSSP) and Enhanced Uni-
`versal Synchronous Asynchronous Receiver Transmitter (EU-
`SART) modules [14]) and using General-Purpose Input Out-
`put (GPIO) pins with complete interface implementation by
`microcontroller software (i.e. bit-bang implementation). For
`implementing the interfaces was partially used PIC18 MPLab
`library from Microchip [15].
`The structure of the developed testbed is presented in Fig. 4.
`As revealed in Fig. 4, the developed testbed was using current
`shunt method for consumed energy measurement [16]. The
`consumed by testbed current, that was later used to estimate
`the consumed energy (see (1)), was estimated by measuring
`the voltage drop (Vshunt) over measurement period from T0
`to Tmeas on known shunt resistor (Rshunt = 4.7Ohm) using
`the oscilloscope. For each implementation option and each
`ten measurements for Econs were made
`interface at
`least
`
`Authorized licensed use limited to: Irell & Manella LLP. Downloaded on March 28,2023 at 00:46:30 UTC from IEEE Xplore. Restrictions apply.
`
`Netlist Ex 2052
`Samsung v Netlist
`IPR2022-00996
`
`

`

`Parameters
`
`Number of lines:
`Data transfer:
`Communication model:
`Number of master devices:
`Number of slave devices:
`Maximum data rate, Mbit/s:
`Interface features:
`Robust communication mechanisms:
`Device identification mechanisms:
`Overhead per packet, bits:
`
`a using 7-bit addresses
`b using 10-bit addresses
`
`TABLE I: Comparison of serial interfaces
`
`UART
`
`2
`asynchronous duplex
`peer-to-peer
`-
`-
`50c
`
`parity checkf
`-
`2 per byte
`
`c according to [9]
`d according to [12]
`
`SPI(4 wires)
`
`3 + 1 per slave
`synchronous duplex
`master-to-slave
`multiple
`multiple
`70d
`
`-
`-
`0
`
`I 2C
`
`3 with pull-ups
`synchronous simplex
`master-to-slave
`multiple
`119aor 1024b
`3.4e
`
`acknowledgment
`address
`11 + 1 per byte
`
`e according to [11]
`f optional
`
`TABLE II: General testbed and experiment parameters
`
`Parameters
`
`rate,
`
`UART
`HW SW
`Required testbed resourcesa
`1215
`Flash, bytes (TX)
`1151
`281
`RAM, bytes (TX)
`274
`1187
`Flash, bytes (RX)
`1227
`282
`RAM, bytes (RX)
`291
`Experiment parameters and settings
`Pull-up resistors, kOhm
`-
`-
`Data rate, kbit/sb
`15.71
`15.82
`125
`65
`Maximum data
`kbit/sc
`a using nominal optimization settings of IDE
`b the data rate for interface during experiment
`c using 8MHz core clock frequency and 3.0 V supply voltage
`
`SPI
`HW SW
`
`I 2C
`HW SW
`
`1132
`272
`1045
`288
`
`-
`15.66
`166.7
`
`1261
`281
`1010
`278
`
`-
`15.84
`74.6
`
`1122
`274
`1125
`290
`
`100
`15.94
`222.2
`
`1395
`281
`1190
`278
`
`100
`15.8
`38
`
`(cid:2) Tmeas
`
`Econs =
`
`and the average value (Eexp) for all the measurements has
`been calculated. During all the measurements, the testbed was
`supplied from the same laboratory DC power source with 3V
`voltage (Vcc) and nominal clock frequency of microcontrollers
`in active mode both for RX and TX boards was 8 MHz.
`The maximum error for the measurements is estimated to be
`in the order 5 µs for time and 4% for energy consumption
`measurements.
`
`Vcc · Vshunt(t)
`Rshunt
`
`dt
`
`(1)
`
`T0
`The energy consumption measurements have been done in
`three stages. First of all, energy consumption for TX and RX
`boards has been measured when the serial interfaces on both
`boards have been initialized, but no transmission was ongoing.
`The second and third measurements were made when the
`transmission was ongoing: during the second measurement
`the RX board was selected as data target, during the third
`measurement - RX slave board was not selected as target (for
`UART third measurement was not made as it is impossible to
`deselect the RX). The energy consumption during the second
`and third stages has been measured for two cases: single byte
`transmission and transmission of nine data bytes (T0 - start of
`packet transmission; Tmeas - end of packet transmission) in
`
`Fig. 4: Experiment setup
`
`single packet (all required service operations, e.g., start/stop
`bits transmission or slave address transmission for I2C are also
`accounted).
`To have the possibility to compare the results of the
`measurements, all
`interfaces have been tested at
`the data
`rate of around 15.8kbit/s (the measured real-life data rate
`(estimated basing on the time for transmitting single bit in
`the middle of data byte transmission, all service operations
`excluded) is presented in Table II). The main reasons for the
`difference between the data rates for various interfaces and
`implementations are different mechanisms for clock generation
`for various interfaces and microcontroller clock instability.
`During second and third stages,
`the transmitted data (in
`hexadecimal) was 0xAA for single byte transmission and
`0x000103070F1F3F7FFF for nine-byte data packet.
`
`IV. RESULTS
`testbed and experiment parameters and the
`The general
`results of energy consumption measurements for implemented
`interfaces are summarized in Tables II and III respectively.
`To be able to compare the results for different interfaces and
`their implementation methods, the values of energy for data
`
`Authorized licensed use limited to: Irell & Manella LLP. Downloaded on March 28,2023 at 00:46:30 UTC from IEEE Xplore. Restrictions apply.
`
`Netlist Ex 2052
`Samsung v Netlist
`IPR2022-00996
`
`

`

`transmission have been normalized using Eq. 2 to the common
`data rate value (15.8 kbit/s). In Eq. 2 Datarateexp stands for
`the experimentally defined data rate in kbit/s (see Table II),
`Eexp - for the experimentally measured energy consumption
`for transferring one or nine data bytes respectively(see Table
`III).
`
`(cid:3)
`
`Enorm =
`
`15.8 − Datarateexp
`15.8
`
`(cid:4)
`+ 1
`
`· Eexp
`
`(2)
`
`V. DISCUSSION AND CONCLUSION
`
`In this paper we have discussed and evaluated the most
`interfaces (namely I2C, SPI and
`widespread digital serial
`UART), that are used by different embedded systems and have
`evaluated real-life energy consumption for different scenarios
`and implementations of those interfaces.
`As the data presented in the paper reveal, of all tested
`interfaces SPI had the lowest power consumption both with
`inactive interface and during data exchange for the case, when
`all interfaces were using the same data rate. Nonetheless, SPI
`interface does not support any standardized mechanism for
`device identification or robust communication implementation.
`The presented data reveal that SPI slave receiver implementa-
`tion required the minimum amount of resources comparing
`to other tested interfaces, which is caused by very simple
`data format in SPI. Also this was the reason, why complete
`software SPI implementation allowed to get maximum data
`rate among all tested interfaces.
`The interface based on UART required at least 22-23% more
`energy for transferring the same amount of data comparing to
`SPI. This is mainly caused by the necessity to have some
`overhead for communication via UART - namely start and
`stop bits for each byte. The implementation of UART interface
`for slave device required more resources than implementation
`of SPI interface due to higher complicity for data receival
`procedure without any specific clock signal availability. Unfor-
`tunately, the most widespread UART-based interfaces support
`only peer-to-peer connection, which makes it impossible to
`connect multiple UART devices to single UART port of
`the microcontroller. The UART interface usually can use a
`parity check bit for each data byte, but has no standardized
`mechanisms for device identification support.
`The single byte transmission using I2C interface required
`more than two times more energy comparing to SPI, although
`for longer packets the efficiency of I2C interface increases
`and for packets over 9 bytes it becomes more efficient than
`UART. The reason for it is the necessity for I2C to start each
`data packet with 1 byte containing slave device’s address. The
`software implementation for I2C interface required rather high
`resource consumption, which is caused by high complicity
`for I2C communication procedures (e.g., use of start, restart
`and stop bits and acknowledgment). This is also the reason
`why complete software implementation for I2C interface had
`almost two times lower maximum achievable data rate than
`the other interfaces. Surprisingly, hardware implementation
`for I2C had the maximum achievable data rate of all tested
`interfaces, which should be caused by the features of interface
`hardware modules implementation in PIC18 microcontrollers.
`The I2C has the support for device identification (using 7-
`byte addresses) and uses per-byte acknowledgments during
`data transmission.
`The data presented in paper reveal, that although implemen-
`tation of serial interface in software requires approximately
`same amount of resources as implementation in hardware,
`it causes significantly higher power consumption (20%-300%
`
`As the presented data (see Table II) reveal, surprisingly,
`the complete interface implementation in software required
`almost same amount of resources as the implementation of
`interfaces using hardware modules. Besides, for some cases
`(slave receivers for UART and SPI), the resource requirement
`for bit-bang implementation was even slightly lower than for
`hardware-based one. The programs that have been used for
`estimating those resource requirements had the same func-
`tionality and used Microchip PIC18 MPLab library [15] for
`implementing interfaces using hardware modules and proper
`assembly-based code for implementing the interface in soft-
`ware. Nonetheless, as revealed in Table II,
`the maximum
`achieved data rate (i.e. the maximum achieved data rate for
`error-less communication between the boards working at 8
`MHz microcontroller clock frequency) for hardware imple-
`mentation is significantly higher than the one for complete
`software implementation.
`the energy consumption for data
`Table III reveals,
`that
`transmission over the interface implemented in software is
`significantly higher than the one for the same interface im-
`plemented using inbuilt microcontroller hardware modules.
`This is especially notable for SPI interface, for which the
`difference in energy consumption is almost 3 times. The
`major reason for this difference is the possibility for using
`for hardware-implemented interfaces the interrupts on data
`transmission/receival complete. Those allow to keep the mi-
`crocontroller in sleep mode most of the time thus significantly
`lowering the power consumption. This is especially important
`for low-duty cycle applications such as e.g., many WSNs.
`As one can see from Table III, the most efficient interface
`from the point of energy consumption was SPI. This is not
`surprising, as SPI is the only interface among tested ones,
`that does not have any overhead data transmission. If to
`compare the measurements for software implementation (this
`is the most rational as interface implementations in hardware
`required to use different sleep modes due to different sets
`of required peripherals such as timers and clock systems),
`the UART required 22.5% and I2C - 26.6% more energy for
`transferring 9 data bytes. Also, as can be seen from presented
`data, the difference in energy consumption between UART
`and I2C for single byte transmission was around two times
`(as I2C had to transmit one extra byte with slave address) and
`for 9-byte transmission this difference almost vanished and
`was only 4%. This allows to expect that for packets with over
`10 data bytes the I2C interface will be more energy-effecient
`than UART.
`
`Authorized licensed use limited to: Irell & Manella LLP. Downloaded on March 28,2023 at 00:46:30 UTC from IEEE Xplore. Restrictions apply.
`
`Netlist Ex 2052
`Samsung v Netlist
`IPR2022-00996
`
`

`

`TABLE III: Energy consumption for evaluated serial interfaces
`
`Parameters
`
`Interface implementation in hardware
`Power consumption with inactive interface, mW
`Communication:
`Required time, ms
`Required energy, µJ
`Normalized energya,µJ
`
`Interface implementation in software
`Power consumption with inactive interface, mW
`Communication:
`Required time, ms
`Required energy, µJ
`Normalized energya,µJ
`
`a see Eq. 2
`
`UART
`
`SPI
`
`I 2C
`
`1 byte
`
`9 bytes
`
`RX selected
`1 byte
`9 bytes
`
`RX unselected
`1 byte
`9 bytes
`
`RX selected
`1 byte
`9 bytes
`
`RX unselected
`1 byte
`9 bytes
`
`6.06
`
`0.68
`7.31
`7.27
`
`6.06
`
`5.78
`53.31
`53.01
`
`4.66
`
`0.52
`2.5
`2.48
`
`4.66
`
`4.6
`23.53
`23.33
`
`4.66
`
`0.52
`2.42
`2.4
`
`4.66
`
`4.6
`22.52
`22.32
`
`6.13
`
`1.26
`8.11
`8.19
`
`6.13
`
`5.91
`31.71
`31.99
`
`6.13
`
`1.26
`8.01
`8.08
`
`6.13
`
`5.91
`30.23
`30.5
`
`11.04
`
`11.04
`
`11.17
`
`11.17
`
`11.17
`
`11.17
`
`11.23
`
`11.23
`
`11.23
`
`11.23
`
`0.64
`9.00
`9.01
`
`5.72
`80.06
`80.16
`
`0.51
`7.26
`7.28
`
`4.62
`65.25
`65.42
`
`0.51
`7.26
`7.28
`
`4.61
`65.25
`65.42
`
`1.23
`17.45
`17.45
`
`5.85
`82.84
`82.84
`
`1.23
`17.44
`17.44
`
`5.86
`83.6
`83.6
`
`depending on the interface) and allows to achieve lower max-
`imum communication data rate (2-6 times lower comparing to
`implementation in hardware).
`Although all our experiments have been executed using
`only one hardware platform (namely PIC18F45K20 microcon-
`trollers), we assume that obtained results can be expanded
`to some extent also for other hardware platforms. We be-
`lieve that the data presented in the paper provide valuable
`information, that can help to choose the most energy efficient
`communication interface and its implementation method to be
`used in the most energy-critical applications. One of major
`target application groups, that can benefit from the presented
`data utilization, are the Wireless Sensor Networks. Indeed,
`as has been revealed e.g., in [7], nowadays the SPI and I2C
`interfaces are very widespread among different digital sensors
`(e.g., among the available digital temperature sensors I2C bus
`is used by 57% and SPI - by 10%). Therefore, for obtaining
`high energy efficiency e.g., in WSN one should consider not
`only the actual parameters of a sensor, but its communication
`interface as well and the data presented in the paper provide
`the background information for it.
`In the current paper we have evaluated the serial interfaces
`for one particular data rate and estimated the maximum data
`rates that can be achieved for various implementation options.
`In future we are planning to make similar tests for other
`hardware platforms and other data rates which would allow to
`understand better the influence of serial interface’s parameters
`on the power consumption of the whole system. This will
`allow to develop, in future, a model that could predict the
`power consumption for communication over serial interfaces
`between various components of a portable electronic device
`for specified interface settings and environment conditions.
`
`ACKNOWLEDGMENT
`The authors wish to thank all parties who have financially
`supported this study. This work has been supported by Euro-
`pean Regional Development Fund, Council of Oulu Region,
`the Sub-regions of Ylivieska and Nivala-Haapajarvi, Town of
`Ylivieska and Kerttu Saalasti Foundation.
`
`REFERENCES
`[1] B. Munsey. (2011, Aug.) New developments in battery design and trends.
`House of Batteries. [Online]. Available: http://www.houseofbatteries.
`com/documents/New%20Chemistries%20April%202010%20V2.pdf
`[2] K. Mikhaylov, J. Tervonen, and D. Fadeev, “Energy efficiency aware
`application development using programmable commercial low-power
`embedded systems processors,” in Embedded Systems - Theory and
`Design Methodology, K. Tanaka, Ed. Rijeka, Croatia: InTech, 2012,
`pp. 407–430.
`[3] G. Mathur, P.Desnoyers, D. Ganesan, and P. Shenoy, “Ultra-low power
`data storage for sensor networks,” in Proc. IPSN 2006, 2006, pp. 374
`–381.
`[4] M. Kuorilehto, M. Kohvakka, J. Suhonen, P. Hamalainen, M. Han-
`nikainen, and T. Hamalainen, Ultra-Low Energy Wireless Sensor Net-
`works in Practice: Theory, Realization and Deployment. Hoboken, NJ:
`John Wiley & Sons, 2007.
`[5] N. Sklavos and K. Touliou, “A system-level analysis of power consump-
`tion and optimizations in 3G mobile devices,” in Proc. NTMS 2007, May
`2007, pp. 225–235.
`[6] K. Mikhaylov, J. Jamsa, M. Luimula, J. Tervonen, and V. Autio,
`“Intelligent sensor interfaces and data format,” in Intelligent Sensor
`Networks: Across Sensing, Signal Processing, and Machine Learning,
`F. Hu and Q. Hao, Eds. London, UK: Taylor and Francis LLC, CRC
`Press, to appear in 2012.
`[7] K. Mikhaylov, T. Pitkaaho, and J. Tervonen, “Plug-and-play mechanism
`for plain transducers with digital interfaces attached to wireless sensor
`network nodes,” submitted for publication.
`[8] G. Gridling and B. Weiss, Introduction to Microcontrollers. Vienna,
`Austria: Vienna University of Technology, 2007.
`[9] H. Yuan, J. Yang, and P. Pan, “Optimized design of UART IP soft core
`based on DMA mode,” in Proc. ICIEA 2010, June 2010, pp. 1907 –1910.
`[10] SPI Block Guide v. 03.06, Motorola Semiconductor Products Inc. Std.
`S12SPIV3/D, 2003.
`[11] I2C - bus specification and user manual rev.03, NXP Semiconductors
`Std. UM10 204, 2007.
`[12] D. Johnson, “Implementing serial bus interfaces using general purpose
`digital instrumentation,” IEEE Instrumentation Measurement Magazine,
`vol. 13, no. 4, pp. 8 –13, August 2010.
`[13] “PICkit 3 Programmer/Debugger Users Guide,” Microchip Technology
`Inc.