ITU-R

A telecommunications regulatory body that regulates or decides how different parts of the RF spectrum may be used. Countries can also have own regulatory bodies that regulate the spectrum within the country. Maintains spectrum in three different regions:

Region 1: Europe, Africa, Northern Asia
Region 2: North and South America
Region 3: Southern Asia and Australiasa

Most bands in the RF spectrum are tightly regulated and require a license. Using a frequency in a licensed range requires an organization to submit an application to the regulatory body.

ITU-R allocated two two ranges for Industrial, Scientific and Medical (ISM) use:

2.400 to 2500 GHz
5.725 to 5.825 GHz

ISM bands are unlicensed and anyone can use them.

Unlicensed bands are more vulnerable to interference and noise due to them being more accessible.

FCC

Federal Communications Commission (FCC) regulates RF frequencies, channels and transmission power within the US but other countries may also follow the rules of the FCC. FCC has allocated Unlicensed National Information Infrastructure (U-NII) in addition to the ISM band. Consists of four bands in the 5 GHz band.

U-NII-1 5.15 to 5.25 GHz
U-NII-2 5.25 to 5.35 GHz
U-NII-2 Extended 5.47 to 5.725 GHz
U-NII-3 5.725 to 5.825 GHz (also allocated as ISM)

FCC must approve transmitting equipment before it can be sold. There are strict limits on EIRP.

Be aware of the EIRP limits. An antenna with too much gain could push you over the EIRP limits for a specific transmitter.

2.4 GHz transmitters can have a transmit power of max 30 dBm and EIRP of 36 dBm. That means that the antenna gain is assumed to be 6 dBi.

Point-to-multipoint links – The transmitted signal propagates in all directions. The 1:1 rule specifies that for each dBm removed from the transmitter one dBi of antenna gain can be added if the EIRP does not go above 36 dBm.

Point-to-point links – The 3:1 rule specifies that for each dBm removed from the transmitter, 3 dBm gain can be added to the antenna. EIRP can exceed 36 dBm but not 56 dBm.

Band Allowed Use Transmitter Max EIRP Max
U-NII-1 Indoor only 17 dBm (50 mW) 23 dBm
U-NII-2 Indoor or outdoor 24 dBm (250 mW) 30 dBm
U-NII-2 Extended Indoor or outdoor 24 dBm (250 mW) 30 dBm
U-NII-3 Indoor or outdoor 30 dBm (1 W) 36 dBm

Transmitters in 2.4 GHz or 5 GHz band must endure interference caused by other transmitters. However in the U-NII-2 and U-NII-2 Extended bands, transmitters must move to another frequency if an approved device such as a weather radar or military device is detected. This is known as Dynamic Frequency Selection (DFS).

ETSI

In Europe and in several other countries radio transmitter usage is regulated by European Telecommunications Standards Institute (ETSI).

Band Allowed Use EIRP Max
2.4 GHz ISM Indoor or outdoor 20 dBm
U-NII-1 Indoor only 23 dBm
U-NII-2 Indoor only 23 dBm
U-NII-2 Extended Indoor or outdoor 30 dBm
U-NII-3 Licensed N/A

ETSI regulation also includes DFS but note that the U-NII-3 band is licensed.

Other Regulatory Bodies

Naturally there are regulations outside of America and Europe as well. Does that mean that every country has their own regulations though? Some countries adhere to parts or all of the regulations of a larger, more established regulatory body. Countries with a common set of RF regulations are known as a regulatory domain.

Devices compatible with American regulatory domain can be used in Canada, many Latin and South American countries as well as the Phillipines.

Cisco manufactures wireless devices for use in at least 13 regulatory domains. Operation is identical in all domains but frequency ranges, channels and maximum transmit power can differ.

IEEE

Institute of Electric and Electronic Engineers (IEEE) is responsible for defining wireless standards and many others where wireless is defined in the 802.11 standard.

As the standard evolves, admendments are added to the original standard such as 802.11a, 802.11b, 802.11g, 802.11n and so on. Periodically the whole 802.11 standard gets revised and the admendments get “rolled up” into the main standard. The names are still used to refer to the technologies though such as 802.11n.

802.11 Channel Use

The 2.4 GHz band is divided into 14 channels, numbered 1 through 14. The channels are spaced 5 MHz apart except for channel 14. DSSS or OFDM can be used in the 2.4 GHz band. DSSS channels are 22 MHz wide and OFDM channels are 20 MHz wide. FCC limits the band to channels 1 to 11, while ETSI allows 1 to 13 and Japan permits all 14 but channel 14 has some restrictions.

The only non-overlapping channels are channels 1, 6 and 11.

Channels in the 5 GHz Bands

Only OFDM is allowed in the U-NII bands. There’s up to 23 non-overlapping channels available in the U-NII band which is a lot more than what is available in the 2.4 GHz band.

802.11 Standards

802.11-1997

The original standard that was ratified in 1997 and only supporting FHSS and DSSS, for use in the 2.4 GHz band.

Band Transmission Type Modulation Data Rate
2.4 GHz FHSS 1, 2 Mbps
2.4 GHz DSSS DBPSK 1 Mbps
2.4 GHz DSSS DQPSK 2 Mbps

802.11b

Introduced in 1999 and offered data rates of 5.5 and 11 Mbps through the use of CCK. Backward compatible since it was based on DSSS and operating in 2.4 GHz band. 1, 2, 5.5 or 11 Mbps could be used by device by changing modulation and coding schemes.

Band Transmission Type Modulation Data Rate
2.4 GHz DSSS DQPSK with CCK 5.5 Mbps
2.4 GHz DSSS DQPSK with CCK 11 Mbps

802.11g

Introduced in 2003 and based on OFDM. Commonly called Extended Rate PHY (ERP) or ERP-OFDM.

Band Transmission Type Modulation Data Rate
2.4 GHz ERP-OFDM BPSK 1/2 6 Mbps
2.4 GHz ERP-OFDM BPSK 3/4 9 Mbps
2.4 GHz ERP-OFDM QPSK 1/2 12 Mbps
2.4 GHz ERP-OFDM QPSK 3/4 18 Mbps
2.4 GHz ERP-OFDM 16-QAM 1/2 24 Mbps
2.4 GHz ERP-OFDM 16-QAM 3/4 36 Mbps
2.4 GHz ERP-OFDM 64-QAM 2/3 48 Mbps
2.4 GHz ERP-OFDM 64-QAM 3/4 54 Mbps

Wireless devices can choose data rates of 6, 9, 12, 18, 24, 36, 48, 54 Mbps by changing the modulation scheme. The higher rates can be used when there is an optimal SNR.

Because 802.11g has higher data rates than 802.11b it would be beneficial to use it everywhere but it’s not possible if there are still 802.11b devices. 802.11g uses OFDM and 802.11b uses DSSS which means they are not compatible and don’t understand each other’s RF signals.

802.11g was designed to be backward compatible. 802.11g devices are able to downgrade and understand 802.11b DSSS messages but the reverse is not true, 802.11b devices do not understand 802.11g OFDM messages. If two 802.11g devices are communicating with OFDM, 802.11b devices won’t understand the transmissions, and might interrupt with their own transmissions.

802.11g and 802.11b can coexist on a wireless LAN by using a protection mechanism. Each OFDM transmission is preceded by DSSS flags that 802.11b devices can understand. When the 802.11g device is ready to send data in protection mode, it first sends a Request To Send (RTS) and then a Clear To Send (CTS) using DSSS and a low data rate. 802.11b devices will then know that an OFDM transmission will follow and that they should wait a predefined time until the transmission is complete because the OFDM messages is unintelligble to them.

802.11a

802.11a introduced use of 5 GHz U-NII band for wireless LANs which has more non-overlapping channels and less interference since only one band in U-NII is used for ISM. In the 2.4 GHz band there are only 3 non-overlapping channels, limiting the scalability.

802.11a is OFDM only. The modulation schemes and data rates are identical to 802.11g but with less risk of interference and more room for growth.

Band Transmission Type Modulation Data Rate
5 GHz OFDM BPSK 1/2 6 Mbps
5 GHz OFDM BPSK 3/4 9 Mbps
5 GHz OFDM QPSK 1/2 12 Mbps
5 GHz OFDM QPSK 3/4 18 Mbps
5 GHz OFDM 16-QAM 1/2 24 Mbps
5 GHz OFDM 16-QAM 3/4 36 Mbps
5 GHz OFDM 64-QAM 2/3 48 Mbps
5 GHz OFDM 64-QAM 3/4 54 Mbps

802.11a is not backward compatible and does not support data rates below 6 Mbps or DSSS. One of eight modulation schemes is selected to support data rates of 6, 9, 12, 18, 24, 36, 48 or 54 Mbps.

802.11a is based on OFDM channels that are 20 MHz wide. This is a perfect fit for U-NII bands that have channels that are spaced 20 MHz apart but there is still some overlap on adjacent channels. For that reason it’s recommended for transmitters in the same area to be separated by one channel. If one transmitter uses channel 36 the other should use 44 and not 40 which is the adjacent channel.

802.11a was introduced in 1999, earlier in the same year as 802.11b. Remember that standards are assigned chronologically. OFDM came to 2.4 GHz in 2003 through 802.11g because going to 802.11a required new hardware.

802.11n

Published in 2009 with a theoretical maximum of 600 Mbps transmission. Uses a number of techniques known as High Throughput (HT) that can be applied to either the 2.4 GHz or 5 GHz band. Backward compatible with OFDM used in 802.11g and 802.11a.

Before the 802.11n standard, devices would use a single transmitter and receiver. These components formed one radio, resulting in a single radio chain. This is a Single In Single Out (SISO) system. 802.11n uses multiple radio components, forming multiple radio chains. This means that a 802.11n device can have multiple antennas, multiple transmitters and receivers. This is known as Multiple Input Multiple Output (MIMO).

The number of radio chains available are described with TxR notation where T is the number of transmitters and R the number of receivers. A 2×2 MIMO device has 2 transmitters and 2 receivers while a 2×3 device has 2 transmitters and 3 receivers. A 802.11n device must have at least 2 radio chains (2×2) and a maximum of 4 (4×4).

802.11n uses the following features to improve throughput:

  • Channel aggregation
  • Spatial Multiplexing (SM)
  • MAC layer efficiency

It also uses the following to improve RF reliability:

  • Transmit Beamforming (TxBF)
  • Maximal Ratio Combining (MRC)

Channel Aggregation

OFDM has 48 subcarriers in each 20 MHz channel. 802.11a and 802.11g devices only have one transmitter and receiver and can only operate in one channel at a time.

802.11n has 52 subcarriers and the radio can operate on 20 MHz channel or 40 MHz channel. The use of a 40 MHz channel doubles the throughput.

An aggregated channel must bond two adjacent channels such as channel 36 and 40.

When using 20 MHz channels there’s some quiet space between adjacent channels. When aggregating two adjacent channels this space can be used for subcarriers which is why a 40 MHz channel can have 108 subcarriers even though each 20 MHz channel normally can have 52 subcarriers, which would total to 104 subcarriers if just adding them together.

The number of available channels decrease when using 40 MHz channels compared to 20 MHz channels. When using 40 MHz channels there are 11 non-overlapping channels compared to 23 when using 20 MHz.

In the 2.4 GHz band there are only three non-overlapping channels so it’s not feasible to do channel aggregation there.

Spatial Multiplexing

When using channel aggregation the bandwidth is doubled by doubling the channel width but a single radio chain is still used. Remember that a 802.11n device can have multiple radio chains. Data throughput can be increased if the data can be sent over multiple radio chains, all operating on the same channel, but separated by spatial diversity. This is known as spatial multiplexing.

How is it possible for several radios to transmit on the same channel without interfering with each other? By trying to keep the signals isolated or easily distinguished from one and other… Each radio chain has an antenna and the antenna is spaced some distance apart. The signals arriving at the receiver’s antenna (also spaced apart) will then likely arrive out of phase or with a different amplitude. This is especially true if the signals take a slightly different path and bounce off of different walls etc.

Data can be distributed across the transmitter’s radio chains in a known fashion. Even several independent streams of data can be processed as spatial streams and multiplexed over the radio chains. The receiver must be able to interpret the signals and rebuild the original data streams by reversing the multiplexing of the transmitter (demultiplexing).

This requires quite a bit of digital signal processing at both ends. The throughput of a channel is increased by having several spatial streams. The more spatial streams available, the more data that can be sent over the channel.

Remember how in Chapter 1 it was mentioned that having the signal arrive out of phase would render in a weaker signal? It’s interesting how spatial multiplexing utilizes on this to actually increase the throughput of the channel.

Earlier we used the TxR notation to show how many transmitters and receivers a NIC could support. To show how many spatial streams are supported we add a colon to this notation so that 3×3:2 means that the NIC would have three transmitters, three receivers and two unique spatial streams.

The number of spatial streams supported is not always equal to the number of transmitters or receivers. Each spatial stream is distributed across the radio chains and not neccesarily associated with a specific transmitter or receiver. The number of spatial streams supported is dependant on the processing capacity and the transmitter feature set of the 802.11n device and not on the number of radios.

Ideally two 802.11n devices communicating should support the same number of spatial streams but this is not always the case. Therefore it must be possible to negotiate the number of spatial streams used by informing the other party of your capabilities and selecting the lowest common number of spatial streams supported. It is however possible for a transmitting device to leverage an additional spatial stream to repeat some information for increased redundancy.

MAC Layer Efficiency

802.11n doesn’t only rely on more radio chains to make communication more efficient. It also uses other methods such as the following:

Block Acknowledgement

802.11 specifies that every transmitted frame must be acknowledged. If the frame was unacknowledged it must be resent. Acknowledging every frame is inefficient and uses up airtime.

802.11n allows for sending data in bursts where only one acknowledgement needs to be sent to acknowledge the burst of data and not each individual frame. This means that more airtime can be used for sending data.

Guard Interval

When OFDM symbols are transmitted they can take different paths to reach the receiver. If the two symbols arrive too close together, they can interfere with each other and corrupt the received data. This is known as Intersymbol Interference (ISI). To protect against this, the 802.11 standard requires a Guard Interval (GI), a period of 800 nanoseconds between each OFDM symbol transmitted to protect against ISI.

802.11n allows for a GI of 400 nanoseconds, allowing for OFDM symbols to be transmitted more often and hence increasing throughput by around 10% but at the risk of making data corruption more likely.

Transmit Beamforming

Normally when a transmitter with a single radio chain transmits data, any receivers that are present have an equal opportunity to receive and interpret the data. The transmitter doesn’t do anything to prefere one receiver over another and each receiver is at its mercy of the environment and surrounding counditions to receive at a decent SNR.

With 802.11n the transmitted signal can be customized to prefer one receiver over another. MIMO can be used to send the same signal over multiple antennas to reach specific clients more efficiently.

Normallly when signals travel over different paths, they may arrive delayed and out of phase with each other. This is destructive and leads to a lower SNR and corrupted signal. However with Transmit Beamforming (TxBF) the phase of the signal is altered as it’s fed into the transmitting antenna so that the resulting signals will arrive in phase at a specific receiver. This improves the signal quality as the signals add to each other and leads to a better SNR.

TxBF can use explicit feedback from an 802.11n device at the far end and adjust the transmitted signal phase. The TxBF device kan keep a table with information for each receiving device so that it knows how to alter the phase appropiately for each receiving device. The feedback mechanism is complex to implement and for compatibility issues between vendors and mechanisms, no feedback mechanism has been implemented.

For Cisco devices, ClientLink can be used which does not rely on a feedback mechanism. Based on the data that is received from the far end device, the phase values can be calculated and performed on data transmissions destined for that device. This also means that this works for 802.11a/g devices and not only for 802.11n devices.

Maximal-Ratio Combining

When a signal arrives at the receiver it may look little like the original signal and be degraded or distorted. MIMO allows for transmitting the signal over multiple antennas so that the receiver can receive multiple copies of the same signal. One signal may be better than the others, at least for a period in time. 802.11n allows for Maximal-Ratio Combining (MRC) to take the several copies and produce one signal which represents the best version at any given time. The reconstructed signal will then have an improved SNR and receiver sensitivity.

802.11n Modulation and Coding Schemes

802.11n is backwards compatible with 802.11a/g and can use the same 8 schemes for OFDM using BPSK, QPSK and QAM. 802.11n allows for 32 coding schemes, 8 per spatial stream and each has an index in the Modulation and Coding Scheme (MCS). Channel aggregation and Guard Interval increase the number of possible rates as well and 802.11n allows for a total of 128 possible data rates.

802.11ac

People just can’t seem to get enough of bandwidth so something had to be done after 802.11n to allow for more bandwidth and higher scalability. That is why 802.11ac saw the light in 2013. The goal was to have wireless on par with Gigabit Ethernet and provide a set of capabilities known as Very High Throughput (VHT) supporting speeds of up to 6.93 Gbps.

Here are some of the technologies used to improve on the data rates and scalability of 802.11n. You will recognize many from 802.11n.

Better Channel Aggregation

40 MHz channels can be bonded into 80 MHz or even 160 MHz channels. 802.11ac is only supported in the 5 GHz band because the extensive use of channels.

More Dense Modulation

802.11n supported 64-QAM allowing for modulating the RF signal in 64 different ways, 802.11ac allows for 256-QAM. More data can then be input at a time and throughput is boosted.

MAC Layer Efficiency

Even more data can be aggregated, lowering the overhead and freeing up airtime.

Explicit TxBF

A single feedback method is supported in 802.11ac.

Scalable MIMO

Allows for up to 8 spatial streams.

Multi-user MIMO

MU-MIMO allows for and 802.11ac AP to send multiple frames to multiple receiving devices simultaneously.

Robust Channel Aggregation

20, 40, 80 or 160 MHz channel width supported. There are eleven 40 MHz channels, five 80 MHz channels and two 160 MHz channels available. 802.11ac allows for a different channel width on a frame by frame basis. Wide channels can then overlap within the band as long as they don’t transmit simultaneously.

Not all transmissions will require wide channels, so the channel space can be negotiated for each frame. If a wide channel is needed and is available, it can be used. If some fraction of the channel is already in use, the remainder can be claimed for transmission.

APs are normally placed on non-overlapping channels but when wide channels like 80 MHz or 160 MHz are used, it’s likely they will overlap. Either AP can then claim the wide channel as long as the other AP is not already transmitting there.

Contention for the wide channel is handled by using RTS and CTS frames. When an AP is ready to transmit it sends RTS frame on the primary 20 MHz channel and then replicates the frame to all other 20 MHz channels that are components of the wider channel. The AP is then requesting to use the entire wide channel for one frame. The intended receiver will check to see if the full channel is free, then reply with CTS frames on each free component 20 MHz channel. The AP will then know which parts of the channel are free and transmit there.

Dense Modulation

802.11ac devices can use 256-QAM modulation which gives about 25% more throughput than the best modulation in 802.11n, 64-QAM. 256-QAM requires higher SNR and hence the client being closer to the AP.

There are 10 MCS values available between the client and AP in 802.11ac as opposed to 32 or more in 802.11n. With 802.11ac the number of MCS choices are only tied to the modulation and coding schemes and not the number of spatial streams or the channel width, as in 802.11n. There are more than 10 data rates available in 802.11ac, but not more than 10 MCS choices. 802.11ac has a bit more flexibility where parameters such as spatial streams can be selected more independently from the MCS choice.

MCS Index Modulation and Coding Scheme
0 BPSK 1/2
1 QPSK 1/2
2 QPSK 3/4
3 16-QAM 1/2
4 16-QAM 3/4
5 64-QAM 2/3
6 64-QAM 3/4
7 64-QAM 5/6
8 256-QAM 3/4
9 256-QAM 5/6

MAC Layer Efficiency

Frame aggregation was introduced in 802.11n. An aggregated frame is built by placing multiple payloads, known as MAC Service Data Units (MSDU) inside a 802.11 frame known as PLCP Service Data Unit (PSDU), with one header and trailer which will be transmitted over the air.

Each individual frame (MSDU) can be a maximum of 2304 bytes. In a wired network the frame size would be limited by the MTU but in wireless networks , devices contend for air time and the frame size is bounded by a limit of about 5.5 ms. In 802.11n 64 kB can be transmitted but 802.11ac can transmit 4.5 MB in the same interval.

In 802.11ac every frame is expected to be an aggregate so there is no individual marking of if a frame is an aggregate or not.

802.11ac also supports GI of 400 or 800 ns.

Explicit Transmit Beamforming

Transmit beamforming is an attempt to “focus” or direct a transmission toward a specific client by adjusting the phase of several spatial streams. Transmit beamforming can either be explicit, requiring feedback or passive where the AP infers information about the client from received signals. In 802.11n, explicit TxBF was never implemented.

802.11ac uses only one method for TxBF and it’s known as Null Data Packet (NDP). The AP will first send an NDP announcement frame to identify itself and any other 802.11ac clients present within range. Interested clients can then respond and others will ignore the frame. The AP will then send an NDP frame as a way to “sound” the channel. Clients receiving the NDP can then compute a matrix of information about the channel conditions and how the NDP was received and return the matrix to the AP. The AP will then be able to make beamforming adjustments that are customized for each 802.11ac client as it transmits a frame.

Scalable MIMO

802.11n supports four spatial streams and 802.11ac supports eight spatial streams.

To create multiple paths, several radio chains and antennas are needed and this can challenging, especially for mobile devices, since there is limited space and limited power available.

Multi-User MIMO

802.11n supports MIMO using multiple spatial streams simultaneously but data transfers are dedicated between one wireless user and an AP. In 802.11ac the AP can send data in the downstream direction to multiple users simultaneously (across several spatial streams). There must exist multipath conditions for this to work obviously and explicit transmit beamforming comes in handy since this enables the AP to tailor the transmissions to each specific client. MU-MIMO is not available in the upstream direction from one wireless client toward other devices.

MU-MIMO puts a heavy signal processing burden **on the transmitter to **multiplex the wireless frames across the spatial streams.

802.11ac implementation

802.11ac is available in two waves where the first wave had most of the features but with performance limits that were reasonable for that generation of hardware. Wave 2 arrived in 2016 and the time betwen wave 1 and 2 gave the hardware developers time to produce more advanced products. The following table lists the features and performance of each wave as tested by the Wi-Fi Alliance.

Feature Wave 1 Wave 2
Maximum channel width 80 MHz 160 MHz
Maximum spatial streams 3 4
Maximum modulation 256-QAM (optional) 256-QAM
MU-MIMO No Yes
Maximum typical data speed tested 1.3 Gbps 2.6 Gbps

Noticed how only four spatial streams are supported and a max rate of 2.6 Gbps? There is still ongoing work to support more spatial streams and higher data rates in 802.11ac but this is very complex to accomplish.

802.11 in Other Frequency Bands

802.11 is not only available in 2.4 GHz and 5 GHz bands. Here are some other use cases:

802.11ad – Multi-Gigabit technology that allows devices to operate in the unlicensed 60 GHz band. With such a high frequency, signals tend to propagate less through physical objects. This leads to higher data rates but reduced range. Suitable for very high speed wireless links within a room.

802.11af – Allows unlicensed 802.11 operation in the spectrum that was historically known as Television White Space (TVWS), between 54 and 790 MHz. Several of 802.11ac’s features are implemented such as OFDM, 10 MCS choices, channel aggregation, multiple spatial streams and MU-MIMO. The range is better since the lower frequency penetrates physical objects better.

802.11ah – Allows devices to communicate on frequencies below 1 GHz. The focus is increased range (1 km), lower power consumption, and connectivity to a large number of devices dispersed over the coverage area.

Wi-Fi Alliance

The Wi-Fi Alliance is a non-profit industry association made up of several wireless manufacturers. The Wi-Fi certified program was launched in 2000. To become certified a device is tested against stringent criteria in a testing lab where it’s tested if a device implements a standard correctly. Wi-Fi certified devices are then compatible with each other which was a big problem in the beginning of wireless.

The Wi-Fi alliance have several different programs, some of them listed below:

Wi-Fi Certified n – Correctly implements features such as multiple spatial streams, channel aggregation, block acknowledgement and dual-band operation.

Wi-Fi Certified ac – Implements all of the features of ac including the two waves.

Wi-Fi Direct – Products that can interoperate without the use of an AP for things such as printing, display, and content sharing.

WPA2 – Correctly implements premium personal and enterprise wireless security features.

Protected Management Frames – Premium security to protect Wi-Fi management frames between an AP and wireless devices.

Wi-Fi Protected Setup (WPS) – Easy-to-use initial configuration of wireless security features.

Wi-Fi Multimedia (WMM) – The ability to prioritize and handle various types of traffic with QoS mechanisms.

Voice Personal – Making sure that wireless voice quality is acceptable when using the device.

Voice Enterprise – The ability of Wi-Fi devices to deliver good voice quality, efficient roaming and robust management while voice-capable devices are mobile.

CCNA Wireless – CCNA Wireless Notes Chapter 2
Tagged on:         

Leave a Reply

Your email address will not be published. Required fields are marked *