Power lines were originally devised to transmit electric power from a small number of sources (the generators) to a large

number of sinks (the consumers) in the frequency range of 50-60 Hz. Initially, the first data transmissions over power lines were primarily done only to protect sections of the power distribution system in case of faults. (In fact, power line protection remains one of the primary functions of power line communications.) In such an event, the fast exchange of information is necessary between power plants, substations

Power line communications :

an

and distribution centers so as to minimize their detrimental effects. The logic included the fact that power transmission towers are some of the most robust structures ever built. Thus, from a reliability perspective, any protection signaling scheme would be best served on such networks. Moreover, many remote locations were not hooked up to telephone networks. Thus, it was determined that signaling and exchanging information for power system protection purposes over the existing power lines was the optimal solution. In the past, for protection and telemetering, the largest portions of data were transmitted analogously to voice data using analog telephony. However, in time, automatic telemetering and remote monitoring of status parameters without operators has become increasingly common. So today telephony plays a rather subordinate role to digital communications even on the power lines.

Phone-line networking most commonly referred to as HomePNA is based on the specifications developed by Home Phone Networking Alliance. It is ubiquitousmultiple sockets in each room provide considerable and dispersed DIGITALVISION COMPOSITE: MKC capacityand no new wiring is necessary. The development of the last inch by Home-networking companies in the form wireless network adapters and power-line adapters is gradually leading to widespread home networking; i.e., a wide array of devices connected inside the home in a intra-home network. This in-home networking could transform all power outlets in the household into broadband connections for PCs, telephones and their accessories, as well as overview other enabled electric appliances. Figure 1 illustrates the concept of last inch or in-home networking, while Fig. 2 illustrates the last mile concept. For the last mile access, power line communications is one of the several possible technologies that include cable modem, and different types of Digital subscriber Anindya Majumder lines ( x DSL) and broadband wireless. PLC is not and widely thought to be speed home networking as well as James Caffery, Jr. superior to other techbeing a possible solution for the nologies, nor are the last mile problem. other technologies without problems or The market for power line communiclearly superior to PLC in all respects. cations (PLC) is two-fold: to the home, The major attraction of PLC is that or last mile access; and in the home, or the power lines often already exist. last inch access. According to the Hence, they would be the preferred study, Jumping on the Broadband medium for providing broadband conWagon (a Morgan Stanley Dean Witter nection to rural or remote areas where industry report in April, 2000) power line telephone and cable connections may communications could be better than the not exist. However, it suffers from a other last-inch access technologies such number of problems. as cable, wireless and HomePNA. (DSP). The final piece was the telecommunications market deregulation, first in the US and then in Europe and Asia. All these events have made power line communications a viable technology for high

So why power line communications?

Considering that data transmission over power lines has been around for sometime, one might wonder why it is receiving such renewed attention now? Especially considering, the data rate for protection and telemetering purposes is at most a few kb/sec and is not comparable to the mb/sec data that needs to be supported for multimedia applications. The answer is a combination of effects that took place during the mid thru late 1990s. The most prominent one is, of course, the unparalleled growth of the Internet. This growth was fueled with the technological advancements of very large scale integration (VLSI) and digital signal processing

Residential Gateway

Desktop Personal Computer

PLC Transceiver

PLC Transceiver PLC Transceiver

Fig. 1 The last-inch or in-home networking

Personal Computer

PLC Transceiver

Telephone

Scanner Laptop PLC Transceiver Printer

0278-6648/04/$20.00 2004 IEEE

IEEE POTENTIALS

The power line channel as a transmission medium

First off, the power line carrier was not specifically designed for data transmission and provides a harsh environment for it. Varying impedance, considerable noise that is not white in nature and high levels of frequency-dependent attenuation are the main issues. The channel between any two outlets in a home has the transfer function of an extremely complicated line network. Many stubs have transmission loads of various impedances. Over such a transmission medium, the amplitude and phase response may vary widely with frequency. While the signal may arrive at the receiver with very little loss over some frequencies, it may be completely indistinguishable over other frequencies. Worse, the channel transfer function itself is time varying since plugging in or switching off of devices connected to the network would change the network topology. The location of the transmitter or the receiver (in this case the power outlet) could also have a serious effect on transmission error rates. For example, a receiver close to a noise source would have a poor signal to noise ratio (SNR) compared to one further away from the noise source. The noise sources could be home devices plugged into the network. Just like a wireless channel, signal propagation does not take place between the transmitter and the receiver along a line-of-sight path. As a result, additional echoes must be considered. This echoing occurs because a number of propagation paths exist between the transmitter and the receiver. Reflection of the signal often occurs due to the various impedance mismatches in the electric network. Each multi-path would have a certain weight factor attributed to it to account for the reflection and transmission losses. All reflection and transmission parameters in a power line channel may be assumed to be less than one. The number of dominant multi-paths to be considered (N) is often not more than five or six since additional multi-paths are usually too weak to be of any significance. This is because the more transitions and reflections that occur along a path, the smaller its weighting factor would be. It has been observed from channel measurements that at higher frequencies the channel attenuation increases. Hence, the channel may be described as random and time varying with a frequency dependent signal to noise ratio (SNR) over the transmission bandwidth.
OCTOBER/NOVEMBER 2004

Figure 3 depicts a generalized channel model for the power line physical layer. In Fig. 4, the generalized channel model has been simulated between the frequency ranges 200 Khz to 22 Mhz. The channel exhibits a high level of attenuation with an increasing frequency.

Noise

Noise in power lines is a significant problem for data transmission. This is because it rarely has properties similar to the easily analyzed white Gaussian noise of the receiver front ends. Substation Typical sources of noise Local Loop Distribution Center are brush motors, fluorescent and halogen lamps, Links to switching power supplies To Medium Voltage Network Speech/Data PSTN/WAN and dimmer switches. The noise in power lines can be impulsive or frequency selective in nature Low Voltage Network (230/115V) and, sometimes, both. Several studies on the noise characteristics of power lines have been conducted. Recent papers have discussed not only the type of noise encountered but, also, the distribution of duration, amplitude and inter-arrival time of impulse noise encountered in power lines. According to them, Fig. 2 The last mile broadband the noise in power lines can be classiaccess to homes from the local fied into four categories: distribution center microseconds to milliseconds and has a random inter-arrival time. The psd of such impulse noise may be as much as 50dB above the background noise spectrum. Hence, it is capable of wiping out blocks of data symbols during high data transmission at certain frequencies. It is caused from switching transients in the system network.

with the generators actual supply frequency: This type of impulse noise usually repeats at multiples of the supply frequency of 60/50Hz. It has a short duration of about a few microseconds and a psd that decreases with increasing frequency. The noise is caused from power supplies operating synchronously to the mains frequency. 4. Impulse noise asynchronous with the mains frequency: This is the most detrimental type of noise for data transmission. Its duration varies from a few

Narrow-Band Noise

Asynchronous Impulse Noise

Background Noise

Synchronous Impulse Noise

S(t)

H(t,)

R(t)

Fig. 3 The generalized channel model

Communicating at the PLC Physical Layer

Modulation techniques such as frequency shift keying (FSK), code-division multiple access (CDMA) and orthogonal frequency division multiplexing (OFDM) have been discussed as appropriate modulation schemes for PLC. For low cost, low data rate applications, such as power line protection and telemetering, FSK is seen as a good solution. For data rates up to 1Mbps, the CDMA technique may provide an effective solution. However, for high data applications beyond that, OFDM is the technology of choice for PLC.
5

1. Colored noise: This type of noise has relatively low power spectral density (psd) which decreases with increasing frequency. It is considered to be the sum of all low power noise sources and may be time varying. 2. Narrowband background noise: This noise is mainly due to amplitude modulated sinusoidal signals. This kind of interference is from broadcast stations in the medium and short wave bands. The interference level varies during different times of the day. 3. Impulse noise that is synchronous

Frequency selective fading as experienced by the power line channel severely impairs the capacity of FSK for data rates beyond a few kilobytes per second. A high degree of error control coding would be needed. Combined with the low spectral efficiency of FSK, it would limit the data rate achieved. For CDMA, the signal of each user is spread using a spreading code at the transmitter. It is recovered at the receiver by de-spreading using the same code. CDMA provides robustness against narrowband noise and other forms of interference. Therefore, it seems to be an attractive candidate for PLC. However, in CDMA systems, the processing gain needs to be high to effectively counter narrowband noise and interference from other users. With low processing gain, the robustness against interference and noise is lost and the signal quality may deteriorate to unacceptable levels for all users. The processing gain (PG) of a CDMA system may be expressed as: PG =
Bt Bd

Simulated Power Line Channel

10 0

10 20 30 40 50 60 200 150

les cy Samp Frequen z to 22mhz 0 kh From 20 Fig. 4 Simulated Power line channel model
0 0 200 400

Tim 100 e

50

600

800

1000

1200

(1)

Where, B + denotes the transmission bandwidth and B d denotes the data bandwidth. It is quite evident that for high data rates and for a reasonably high PG, the transmission bandwidth B + would have to be very high. Unfortunately, this is where the problem lies. The fissured spectrum for transmissiondue to frequency selective fadingdoes not provide large contiguous bands for data transmission. Hence, the main advantage of CDMA cannot be fully exploited for PLC. Since the symbol time is inversely proportional to the data rate, as the data rate increases the symbol duration correspondingly decreases. While transmitting over the power line channel at high data rates, the symbol duration is so small that delayed versions of one symboldue to multi-pathsgets smeared over a large number of other symbols. This makes the detection process complicated since it requires complex equalization techniques to counter the inter-symbol interference (ISI). Despite equalization at the receiver, the bit error rate may still be unacceptably high for high data rates over harsh channels with multi-paths. In the case of OFDM modulation, the serial data of a traffic channel is passed through a serial-to-parallel converter. It splits the data into a number

of parallel channels. The data in each channel is applied to a modulator, such that for N channels there are N modulators whose carrier frequencies are f0, f1,., fN-1. These N carriers are referred to as sub-carriers in the literature. This is because they split the work of a single carrier amongst themselves. This scheme offers various advantages. With OFDM, since the data is split among N sub-carriers, each sub-carrier carries 1/Nth of the original data rate. This means that the symbol duration for each sub-carrier increases N times. Moreover, a part of the end of a symbol is appended at its beginning in what forms the cyclic prefix. The length of the cyclic prefix is made longer than the longest delay path. This solves the inter-symbol interference (ISI) problem to a large extent. As a result, a simple linear equalizer may be enough to remove the ISI. Another significant advantage of OFDM, while transmitting over a frequency selective fading channel, is that is allows us to adopt adaptive schemes so we can avoid transmitting at frequencies in deep fade. Sub-carriers in which the signal to noise ratio (SNR) drops below a certain threshold are switched off. Sub-carriers with high SNR are made to carry more bits; i.e., they are modulated to a higher-level constellation. This is known as a bit loading technique and is illustrated in Fig. 5. The application of OFDM with bit loading for a wired channel such as the power line is widely known as Discrete Multi-tone (DMT). Channel estimation is necessary for the bit loading technique to work. The transmitter has to know the noise variance and the attenuation being experienced by each sub-carrier. DMT transceivers use pilot signals for channel estimation. For time varying channels such as the PLC, the pilot signal is repeated periodically for dynamic chan-

nel estimation. In general, bit loading algorithms may be classified as marginadaptive or rate-adaptive algorithms. In margin-adaptive algorithms, the objective is to minimize the bit error rate while keeping the data transmission rate constant. For rate-adaptive algorithms, the data rate is maximized while maintaining a constant error rate. In PLC standards, rate-adaptive algorithms have been adopted. This is because the power line channel can be too harsh at times to guarantee a constant data rate that the margin-adaptive algorithms required. Channel coding plays an important role for maintaining a constant error rate. The input bitstream is encoded using Reed-Solomon coding followed by interleaving and trellis coded modulation. The complete functional block diagram of the DMT transceiver as implemented on the power line physical layer is shown in Fig. 6.

SNR in dB

The PLC Medium Access Control (MAC) Layer

A MAC protocol specifies a resource sharing strategy: the access of multiple users to the network transmission capacity based on a fixed resource sharing protocol. Generally, there are two categories of access schemes: Fixed access Dynamic access Transmissions using fixed access schemes assign to each user a predetermined or fixed channel capacity irrespective of whether the user needs to transmit data at that time. Such schemes are not suitable for bursty traffic such as data transmission that is provided by PLC. Hence, dynamic access is provided for power line communication. Dynamic access protocols may be classified into two separate categories: 1. Contention based protocols: collisions occur. 2. Arbitration protocols: collision free. Contention protocols may not be able to guarantee a quality of service (QOS), especially for time critical applications, since collisions might occur and data might have to be retransmitted. Arbitration based protocols are more capable of guaranteeing a certain QOS. However, contention based protocols may actually provide higher data rates in applications which do not have stringent QOS requirements (e.g., Internet applications). This is because they require much less overhead com-

IEEE POTENTIALS

pared to arbitration protocols (polling, reservation, token passing). Polling and Aloha are the two most studied protocols for medium access. Polling is a primary/secondary access method in which the primary station asks the secondary station if it has any data to send. Aloha is a random access protocol in which a user accesses a channel as soon as it has data to send. The transmitter waits for an acknowledgement from the receiver for a random period of time. It retransmits if it does not receive one. The main disadvantage of Aloha is the low throughput as the load increases as well as the lack of QOS. Arbitration based polling can handle heavy traffic and does provide QOS guarantees. However, polling can be inefficient under light or highly asymmetric traffic patterns or when polling lists need to be updated frequently as network terminals are added or removed. Similarly, token passing schemes (e.g., token ring, token bus) are efficient under heavy symmetric loads. However, they can be expensive to implement and serious problems can arise with lost tokens on noisy unreliable channels such as power lines. Carrier Sense Multiple Access (CSMA) with overload detection has been proposed for PLC. CSMA is a contention based access method in which each station listens to the line before transmitting data. CSMA is efficient under light to medium traffic loads and for many low-duty-cycle bursty terminals (e.g. Internet browsing). The primary advantage of CSMA is its low implementation cost. This is due to the fact that it is the dominant technique in todays wired data networks. Collision detection (CSMA/CD) senses the channel for a collision after transmitting. When it senses a collision, it waits a random amount of time before retransmitting again. CSMA/CD used in Ethernet networks does enhance the performance of CSMA. But on power lines the wide variation of the received signal and noise levels make collision detection difficult and unreliable. An alternative to collision detection that can be easily employed in case of PLC is collision avoidance (CSMA/CA). As in the CSMA/CD method, each device listens to the signal level to determine when the channel is idle. Unlike CSMA/CD, it then waits for a random amount of time before trying to send a packet. Packet size is kept small

due to the PLCs hostile channel characteristics. Though this means more overhead, overall data rate is improved since it means less retransmission. CSMA/CA is the chosen medium access protocol for the Homeplug standard that has been developed for in-home networking using power lines.

Perspective/applications/ standards
The PLC market is expanding dynamically. Some applications and research developments are reported in the International Symposium for Powerline Communications and its Applications (ISPLC) conferences each year. Advanced energy services include applications such as automatic meter reading, programmable controllers and demand supply management. Traditionally, this application area has been pushed by energy companies and related manufacturers. PLC networking in the home would be serving two goals: 1) providing a local home network with the advantages of the power line, and 2) combining access and inhome network capabilities for service and system integration. There are several applications for a PLC network in the home: shared Internet, printers, files, home control, games, distributed video and remote monitoring/security. The key asset here is no new wires. In the United States, the Homeplug Powerline Alliance was founded by Cogency, Conexant, Enikia, Intellon, Netgear, RadioShack Co., Sharp, Panasonic, Cisco systems, Motorola and Texas Instruments, together with several other participants and adopters <http://www.homeplug.org>. The Homeplug Powerline Alliance is a non-profit corporation formed to provide a forum for creating open specifications for high-speed home power line networking products and services. Adopters of the Homeplug 1.0 standard have developed products for inhome networking reaching data rates up to 14Mb/s. The Homeplug standard uses OFDM in a burst mode as the physical layer modulation. The Homeplug technology contains a combination of sophisticated forward error correction (FEC), interleaving, error detection and automatic repeat request (ARQ) to ensure that the channel appears completely reliable to the network layer protocols. The MAC protocol for the Homeplug standard is the CSMA/CA protocol described earlier.

The European Home System (EHS) consortium <www.ehsa.com> defines a bus and a communication protocol for communication between appliances and the central processing unit in the home. The EHS specification, EHS 1.3, covers several medium types to transport control data, power and information. All share the logical link control (LLC) sublayer. For the moment, the supported medium types are Power Line Carrier (230 Vac + data, 2.4 kbps, CSMA/ack, topology free) and Low Speed Twisted Pair (15 VDC, 48 kbps, CSMA/CA, topology free). One major issue under considerable debate for the PLC local loop distribution network is the radiation emission of power lines and its effect on other frequency bands for communication. Another area of concern is security and privacy. The networking signals generated in one home may show up (albeit attenuated) on the power line in another home. This creates urgent concerns about privacy similar to those encountered in wireless systems. Right now, the Homeplug standard uses Data Encryption Standard (DES) encryption technology as a solution.
H(f) Bt

Bits/Sub-Carrier

f1 f2 f3 Frequency Bt

fN

Bits/Sub-Carrier f1 f2 f3 Frequency Fig. 5 The concept of bit loading in which higher bits are allotted to sub-carriers which have higher SNR, and lower number of bits are allotted to sub-carriers experiencing attenuation.

fN

OCTOBER/NOVEMBER 2004

RF TX

DAC

Coding

Interleaving

QAM Mapping

Pilot Insertion

Serial to Parallel

Parallel to Serial

Add Cyclic Extension & Windowing

IFFT (TX) FFT (RX) Remove Cyclic Extension

Decosing

Deinterleaving

QAM Demapping

Channel Correction

Parallel to Serial

Serial to Parallel

Fig. 6 The functional block diagram of DMT as implemented in the PLC physical layer using the Homeplug standard

RF RX

ADC

Timing & Frequency Synchronization

Conclusion
The power line communication field still constitutes an open research area even though successful standards have been adopted in recent years. Many more studies are needed to better understand and improve the performance of power lines for high bit rate applications. DMT (defined earlier, Discrete Multi-tone) technology as adopted by the Homeplug standard can theoretically provide data rates of 100Mb/s. However, products based on the standard only have achieved data rates up to 14Mb/s. The theoretical data rate here indicates the raw physical layer data rate. To protect against the severe noisy conditions and fading in the powerline channel, very high levels of error control coding need to be provided. Hence, the data rate as seen by the upper layers is much less. The efforts of the Homeplug Alliance and home networking technologys growth in the US portend a very bright future for DMT-based PLC home networking.

Read more about it

Klaus Dostert, Powerline Communications, 2001 Prentice Hall, Upper Saddle River, NJ07458. ISBN: 0-13-029342-3 Niovi Pavlidou, A.J. Han Vinck, Javad Yazdani, Power Line Communications: State of the Art and Future Trends, IEEE Communications Magazine, April 2003, pg 34-40. Sobia Baig, N.D. Gohar, A Discrete Multitone Transceiver at the heart of the PHY layer of an In-Home Power Line Communication Local Area Network, IEEE Communications Magazine, April 2003, pg 48-53. Olaf G. Hooijen, Channel Model for the Residential Power Circuit used as a Digital Communications Medium, IEEE Transactions on Electromagnetic Compatibility, Vol. 40, No.4, November 1998. M. Zimmermann, Klaus Dostert, An Analysis of the Broadband Noise Scenario in Powerline networks. Proceedings of International Symposium of Power-line Communications and its applications, 5-7th April, 2000, Limerick, Ireland Halid Hrasnica, Abdelfatteh Haidine, Modeling MAC Layer for Powerline Communications Networks.Internet, Performance and Control network systems, part of SPIEs symposium on Information Technologies, 58th November,2000, Boston, MA, USA/ Majumder, Performance Analysis of M-tone coded frequency hopping signaling over Power line transmission channel, Masters thesis, Southern Illinois University, Carbondale, August 2001. Papaleonidopoulos, I.C.; Capsalis, C.N.; Karagiannopoulos, C.G.; Theodorou, N.J., Statistical analysis and simulation of indoor single-phase low

voltage power-line communication channels on the basis of multi-path propagation, IEEE Transactions on Consumer Electronics , Vol. 49, No.1, Feb.2003.

About the authors

Anindya Majumder obtained his BE in Electronics and Power engineering from Nagpur University, India, in 1998 and Masters degree in Electrical engineering from Southern Illinois University, Carbondale in 2001. During 1998-99, he was employed in BPL Telecom Ltd. at Kolkata, India. He has been pursuing his Ph.D. in Electrical engineering under the guidance of Dr. James Caffery Jr., in the Electrical & Computer Engineering and Computer Science department of University of Cincinnati since September, 2001. His research interests include Wireless Communications, Digital signal processing, Power line Communications and the application of signal processing and data analysis techniques in the area of medical sciences. Dr. James Caffery received his B.S.E.E. (summa cum laude) from Bradley University in 1992. He completed the M.S. and Ph.D. degrees in electrical engineering from the Georgia Institute of Technology in 1993 and 1998, respectively. In 1999, he joined the Department of Electrical and Computer Engineering and Computer Science at the University of Cincinnati where he is currently an assistant professor. His research interests focus on geolocation technology and many aspects of wireless communication systems, including CDMA systems, multiuser transmission/detection and spacetime signal processing.

Dont delay... Renew your membership today.

8

IEEE POTENTIALS

The Private Branch eXchange (also called the PBX or the Private Business Exchange) is a telephone-switching center that is owned by a private business. This switching center is exclusive to one company rather than owned by a common carrier or telephone company. Even so, companies still need their employees to be able call each other at different locations (if need be), call outside phone numbers (the public telephone network or PSTN), and receive calls from the outside. PBXs are distinguished from smaller key systems by the fact that external lines are not normally indicated and selectable from an individual extension. From a users point of view, calls on a key system are made by selecting a line and dialing the external number; calls on a PBX are made by dialing 9 then the external number. Another alternative is to connect all the telephone sets to the PSTN. But the major disadvantage is that every extension requires its own line (usually with a monthly recurring line charge); also, internal calls would be dialed externally and charged for. The main purpose of a PBX is to save the cost of requiring a line for each user to the telephone companys central office. Thus, most local phone companies offer Centrex or Virtual PBX service. That is each extension has a trunk line that is connected to the telephone companys Central Office (CO), where software on the CO switch enables PBXlike functionality. Functionally, the PBX performs three main duties: Establishing connections (circuits) between the telephone sets of two users. Note that fax, modems and many communication devices can often be connected to the PBX (although the PBX may degrade line quality for modems). Therefore, telephone sets are referred to as extensions. Maintaining such connections as long as the users require them. Providing information for the Accounting Department (e.g. metering calls) PBXs offer many capabilities, although each manufacturer may have a different name for each capability. Here is a short list of common capabilities: Direct Dialing (DDD or DDI), also called Direct Inward Dialing (DID) Customized Abbreviated dialing (Speed Dialing) Follow-me

Call forwarding on absence Call forwarding on busy Call transfer Music on hold Automatic ring back Night service Call distribution (Automatic Call Distribution, fixed sequences) The extension interface can be: Proprietary: the manufacturer has defined a protocol. One can only connect the man-

less technology. This can be a cordless system, a small Wireless Local Loop (WLL), Wireless LAN or some cellular technology. The goal is to provide convenience to the workers at a corporate site, so as to achieve in-house mobility and continuous attachment among co-workers and clients Azim A. Samjani, with a unique and constant Rahul R. Shetty contact number. This set up and provides a good amount of scalability, since there is no Rohit George hindrance of laying extra Mathew cables. Also, users can be added without any major modification.
DIGITALVISION COMPOSITE: MKC

Wireless PBX :

using 3G over a wireless LAN

Existing implementations
Wireless PBX exists in a variety of ways. Each way has some advantages and disadvantages. Wired LAN: This is one of the most common approaches used by corporations. Here the PBX switch/machine is connected to a central LAN server. Also, a Base Station (BS) is connected to every PC, via a Universal Service Bus (USB) port. When the PBX wants to contact a particular number, the number is broadcasted on the LAN. If the BS on the USB identifies the number, it responds to the PBX switch. Then the PBX switch identifies the machine, connects it with the user at the calling end and sends voice packets over the network. It is affordable on a LAN, since it provides reasonable bandwidth for such applications. The identification is normally IP-based. The BS then connects to the wireless phone and the user can talk. Typically, a BS can support more than one user. In such cases, it identifies more than one calling number. But when the network is busy or when employees have long calls, then the quality of service is not good. Hence, it is best used for short conversations. Also the computer would be loaded when voice calls are activated. Cordless phones: A cordless phone has a base station and one or more phones connected to it. The base station is connected to the PBX switch with a wire. When a call is to be made, the PBX contacts the correct base station and the call is put through. This is one of the cheapest and easiest systems

ufacturers sets on the PBX Standard interfaces: any device supporting the standard can be connected

Wireless PBX
Wireless PBX (WPBX) is defined as the privately owned and operated wireless telephone equipment, including base stations and handsets, used to connect users to PBX systems or Centrex trunks using unlicensed spectrum. WPBX private wireless networks operate similarly to todays cellular networks except the WPBX networks are privately owned and usually exist within the walls of a business. The WPBX application generally consists of WPBX handsets that use unlicensed spectrum and special WPBX air interfaces to communicate to the nearest private WPBX base station (BS). The base stations are located throughout a work environment to provide the desired coverage. The capacity of the base station may vary depending upon the type of wireless technology used for communication, the backbone network and the devices applications. In wireless PBX, the PBX system is implemented by deploying some wire-

OCTOBER/NOVEMBER 2004

0278-6648/04/$20.00 2004 IEEE

IP PPP LLC MAC Physical MS

Upper Layer Protocols LLC MAC Physical AP

IP LLC MAC Physical LAN Server

3G over WLAN Protocol Stack Fig. 1

to implement. Different cordless technologies can be applied depending upon the type of service desired. Walkie-talkie sets: These are simple and common devices often used to implement wireless PBX. They can be automated systems or there can be a manual switching from PSTN to the internal PBX. These are one of the oldest wireless systems and they have a variety of applications apart from wireless PBX.

Motivation
So what are the problems in the existing wireless PBX? There are many technical issues that are inherent with the technology deployed. They are 1. The company can implement a private protocol or use customized components that are incompatible with other products of the same or different technology. This increases the cost and decreases the scalability. If the compa-

ny wants to interconnect more such components or increase the user base, it is expensive and sometimes impossible (if the technology is old and no more in use). 2. These systems are not standardized and, hence, may at times be insecure. Since it is not a generic technology, companies may have to implement their own security policies. This can prove expensive. Also, since its not tested by others, there is no guarantee the system will be completely secure. 3. When the LAN is used, the network may be heavily loaded when voice packets are passed on to it. This can affect the performance of some data intensive applications. 4. Suppose a company has offices that are geographically dispersed. If an employee moves to another office, his or her number is changed. This creates confusion and, sometimes, connectivity with the person is lost. 5. Since these implementations can be proprietary for a particular corporation, interfacing such a system with clients or other organizations can be difficult at times, and doable only with major modifications.

Scalability. This feature ensures the flexible addition of users without any major changes in the network infrastructure. With a wired PBX, the extra cables required would be result in a installation cost proportional to the amount of infrastructural changes made. Interoperability with other systems. This capability lets the companys wireless PBX network connect to client networks, other commercial networks or the companys own network at some other location. Security. Since we are using a customized system, it is still more vulnerable to attacks. In such cases, the organization has to implement their own security policy. Since this may not be well known, no guarantee about the privacy is justified. Standardized technology. Using widely accepted standard components makes maintenance and upgrades easier. Also, support from different vendors is provided. Always connected. Lastly, an employee is always connected to his or her corporate network irrespective of his or her location.

A cellular based system Wireless PBX using 3G over WLAN

We propose a cellular system based wireless PBX using commercial wireless technical standards. This system supports voice, data and other telephony applications. The data applications involve paging, Short Message Service (SMS), email and such. The telephony applications include caller ID, call forwarding, call waiting and call blocking. For this system, we divide the office into small cells also known as micro cells. For example, every floor in the office could be a cell or a big hall or a shop floor could be a cell. The cells size depends upon the size of the office and may vary from one implementation to another. Every cell has a base station. In turn, every base station is connected to a central switching center, which is the PBX switch. So what cellular technology is incorporated? Any cellular system such as TDMA or CDMA can be used. But we chose a 3G cellular network because: It can operate in an unlicensed 2.4 GHz frequency same as a WLAN. Therefore, no FCC regulations apply and many network products are available which are WLAN enabled. It can support both voice and data due to the considerable increase in bandwidth compared to other cellular

Using third generation (3G) wireless systems over the Wireless LAN (WLAN), these main problems can be solved: Cost of the system. Users subscribCorporate Network-3G over WLAN ing to commercial PSTN cellular operators 3G Switch can use their handsets both outside as well as inside the organizations network. This setup LAN Server eliminates the need for cordless and land line phones. 3G The network comDistribution Network System ponents being used are standardized thus far less costly AP than customized AP ones. Also, the supExternal porting software is Networks BSS readily available. BSS Hence, the maintenance and upgrade of such systems is relatively easy com3G Phones 3G Phones pared to using a proprietary PBX Fig. 2 Network Architecture of 3G over WLAN system.

10

IEEE POTENTIALS

technologies. So it can work on WLAN, which offers a bandwidth of 11 Mbps. These systems are more secure. 3G supports IP and, hence, can interface with other network devices. Thus, the standard 3G protocols are implemented with no modifications. We connect our Mobile Terminal Switching Office (MTSO), which is the 3G switch/wireless PBX switching box, to other commercial MTSOs such as AT&T or Verizon. The precise handshaking protocols and security features are discussed in subsequently. This system is convenient for employees, clients or consultants of this organization since it offers seamless integration into the corporate network without changing instruments. Thus, it keeps them connected at all times. Also, with simple touch procedures, they can switch between networks.

The architecture
This network topology depicts the integration of 3G wireless systems and 802.11 WLAN. The function of the WLAN is to keep the 3G devices connected to the other various networks. They can be LANs, data networks, PSTNs or other 3G networks. The way this network system works is shown in the Fig. 1 protocol stack. Here the LAN server is the interface between the corporate network and the external public network. Actually, there is a small element that directly interacts with the outside world. It is the switch (discussed in the next section). The LAN server handles the data coming
Table 1 ID 989721003 187352920 Cell Phone No. 215-001-009 543-976-275

width of 11 Mbps and can support data rates of 5.5, 2 and 1 Mbps which is still acceptable to the 3G system. The main components in this architecture (see Fig. 2) are the 3G switch/PBX, the LAN Server, the WLAN and the 3G devices. 3G switch/PBX: This is the outermost component of the network. It interfaces with the PSTN and the external commercial 3G network. It accepts input from these sources in the form of voice or IP packets respectively and then switches them to the LAN server. Also, data from the LAN server, which might be voice or data, is sent over to the outbound interface on to their respective networks. LAN server: This server converts the data from the 3G switch to VOIP packets. It also forms the Ethernet frames. When the 3G device gets attached to the corporate network, authentication and other validation procedures are done on this server. A special customized authentication is to be installed which will verify the validity of the user to use the network resources. It also contains the database, which has a map of the telephone numbers to the IP addresses. WLAN: It includes the LAN interface with the access points and the distribution system. It implements the standard CSMA/CD protocol. Here we assume that we have dedicated access points for the 3G devices to carry voice and data. 3G devices: These can be any IPenabled 3G systems such as cell phones, PDAs, etc. But for this discussion, we mainly focus on cell phones.

Office No. 654-129-4533 654-129-4534

IP address 192.168.0.99 192.168.0.100

from the outside network and interprets it into the format of the corporate network. Since 3G devices are IPbased, the external data in converted to Voice Over Internet Protocol (VOIP) packets. In this format, the end devices can access the data, i.e. voice. 3G devices are used as the end system because they can operate on the unlicensed 2.4 GHz frequency used by the WLAN. Also, since they are IP enabled, they are compatible with the TCP/IP network stack. The bandwidth requirements of WLAN and 3G are also compatible, since 802.11b assures a maximum band-

These phones are registered on the corporate network by some authorized entity; e.g., the system administrator. They communicate with the Access Point (AP) over the radio interface.

Basic operations
There are three basic operations that occur very frequently and are important to the operation of 3G phones over the WLAN. They are: 1. Associationgetting connected to the corporate network 2. Registration with Home Networkregistering the cell phone on

the external network 3. Call setupsetting up the calls to and from the cell phone Association: This is a process by which the cell phone gets associated with the Corporate Network (CN). The AP has two channels of communication. One is the control channel and the other one is the data or traffic channel. It always broadcasts control information on the control channel. When the cell phone is switched ON, it accepts the control information. It realizes that it is in the CN and sends a request for association. It sends its ID to the AP. The ID can be an Electronic Serial Number (ESN) or some other identification. The AP forwards the request to the LAN server through the switch. The LAN server interprets the request and checks if there is an entry for the phone; i.e., it checks whether the ID of the phone is present in the database. If its a valid ID, the cell phone is given an IP address. If not, the cell phone is denied association. The database stored on the LAN server contains four entries. They are: 1. Cell phone ID 2. Cell phone number given by the commercial service provider 3. Office number mapped on to this device 4. IP address The database has the format shown in Table 1. The location of the user in the CNwhich cell the user belongs tois also updated. This information is not stored in this database but in a separate management database. The structure of the database can vary and is implementation dependent. We show an example of how such a database may look. Every corporate network can add more details for security and/or can integrate these details in already existing management databases. Registration with the Home Network: After the cell phone gets associated with the CN, it must register the change in its location and IP address to the home network. This has to be done so that calls for the subscriber are directed towards the CN. One very important note, we assume that the 3G switch can interface with the commercial 3G service provider. That is, there is an agreed upon collaboration between the two switching centers to send updates of user information. The switch then makes a session with the external switching center and informs it that their subscriber is in the

OCTOBER/NOVEMBER 2004

11

CN with some IP address. If it is the home agent, then it will update its database as the subscribers foreign network. If it is the foreign network, then it will forward the request to the home network to update its database. Example: Assume that a subscriber is a customer of ABC telecom systems. It gets an IP address from its home network. If the subscriber moves to a foreign network, the home network is made aware of this venture. This foreign network is directly connected to the CNs 3G switch. When this customer gets associated with the CN, then the request is sent to the foreign agent, which in turn forwards to the home agent in the home network the customers new network. When the subscriber leaves the corporate network, the PBX informs the MTSO that the user has changed networks. The MTSO then might restore the state of the user to what he or she held prior to joining the corporate network. Even if the subscriber switches OFF the cell phone, the same procedure is applied since the neither the AP nor the PBX can distinguish this situation: whether the user is outside the network or the cell phone is OFF. Call setup: When a call comes at the inbound interface of the switch, the data is switched over to the LAN server. The LAN server checks the data. This can be voice or an IP packet. From the data, it authenticates the data by checking the destination telephone number. The number is searched in the database with the cell phone number, telephone number or the IP address. If it matches with any one of these, then a VOIP packet is formed and sent back to the switch. The switch then forwards it to the appropriate AP. The AP forwards the packet to the cell phone. If the cell phone is already engaged, then a call waiting or a engaged message is sent back to the caller through the switch. When a user wants to make a call, the voice packet containing the destination details is sent to the LAN server. The user is authenticated (in the case of error the user is ignored). The state of the user is changed to engaged so that no other person can connect to the user. Thus, the user is in a locked state, and only the called person can respond to the user. For valid users, the VOIP packet is converted to voice and sent to the PBX. The PBX then connects the PSTN network to the central office to make a call setup. When the person at the other

3G to PSTN (Outgoing Call) 3G Cell Phone LAN Server PBX/3G Switch

Request for Call Setup Authentication Error: Ignore Forward Request to PBX Forward to LAN Server Authentication Error: Ignore Make Call From PSTN To PSTN

PSTN to 3G (Incoming Call) PSTN PBX Request for Call Setup LAN Server Forward Request Check Data/Integrity Error: Ignore Authentication Error: Ignore To 3G Phone From 3G Phone Authentication Response for Call Setup Fig. 3 Forward Reply Error: Ignore

end replies, he or she is authenticated to confirm his or her identity. Please note that the PSTN network is used to make an outgoing call, since the cost of using a PSTN network is much less than the 3G network. Once the call is setup, the conversion to VOIP packets is done. The sequence of operation for incoming and outgoing calls in shown in Fig. 3. Data interchange: When data arrives from the PBX to the LAN Server the following steps are carried out. 1. The data is checked to distinguish whether it is a voice or an IP packet. Next, the destination ID is extracted. This can be a telephone number or the IP destination address. 2. With this information, the user is authenticated and his or her identity is stored in the database. If it is an invalid user, then the call is ignored. 3. After validation, if it is voice then a VOIP packet is formed. This VOIP packet is then sent to the AP. If it is a data packet then it is simply forwarded to the 3G device. When data arrives at the LAN Server from the WLAN, the following steps are taken. 1. The VOIP packet is opened and the

user is authenticated. If the user is invalid, then the packet is simply dropped. 2. Upon validation, the outgoing network information is retrieved. If it is the PSTN network, then the voice is sent to the PBX, which in turn forwards it to the telephone network. 3. If an IP packet is to be formed, then this packet is simply forwarded to the 3G network through the PBX. These procedures are done at a very high speed, so the user in the corporate network thinks he or she is having a continuous conversation with the person outside the network. Thus, software is needed for authentication and data conversion, to avoid all the processing delays. Also, the network hardware and the LAN server must be fast enough to carry out the processing at high speed. This process is shown in Fig. 4

A billing system
Before the deployment of such a system, one question that arises, How would the cell phone get billed? The answer to this is simple and straightforward. The cell phone has two numbers: one is the number obtained from the commercial service provider, and the other is the office number that

12

IEEE POTENTIALS

is mapped onto it. If the user in not in the CN, then the service provider handles the billing cycle. But if the user is now registered to the CN, then some issues are to be noted. If he or she gets a call on the number provided by the cell phone operator, then the user will be billed by this provider. (He or she can get calls on this number, since the PBX informs the commercial MTSO about the presence of the user in the CN.) But if the user gets a call on his or her office number, then he or she is not billed by the provider, since it is like someone calling him or her on his or her office land line. When the user is making an outgoing call, then the number would be the office number. Hence, the office will be billed for this call, since all outgoing calls from the office PBX is going through the PSTN network.

Drawbacks of the proposed solution

With the proliferation of wireless gadgets and applications in our daily lives, there is an increasing tendency to desire being connected to public and private networks both voice and data. We have proposed a new method of using 3G phones in public and private networks due to their inherent ability to mesh with standard networking technologies. But with this method comes a price tag in terms of monetary resources as well as changes in infrastructure. Apart from these material investments, social issues could arise since we envision a different way of human machine interaction. Technical issues. First of all, it is necessary to use 3G phones. We have designed this system keeping in mind the facilities offered by 3G systems. As of today, 3G phones are in testing mode or being used only by private firms. Hence, it will be awhile before the masses use it. The LAN server is one of the most important components of this system. The efficiency of the entire system depends on it. This server can act as a bottleneck for the entire system, if not well configured and designed to accurately meet desired needs. Normally, simple LAN servers come with dual processors that have more than 2 GHz that is powerful enough to handle voice data operations. But, again, the load on the server depends on the number of users, other applications running on it and also the amount of memory available. More servers can be deployed if

usage warrants it and for load balancing. Also, protocol converters need to be extremely efficient for audible conversation. A one-way delay of more than 100 ms makes it difficult for end users to converse. Hence, protocol converters must not cause a delay of more than 20 ms, since there are other transportation overheads involved. Cost. If the existing wireless LAN is loaded, then some new access points probably would be needed to cater to the needs of the voice calls. But for most of the corporate networks, the existing infrastructure would be suitable for this system. Designing or buying protocol converters for the LAN server can add to the cost. Users need 3G phones, but they would already have them from their commercial service providers. Maintenance of servers, wireless phones and ensuring security policies is an important aspect and must be taken care of for a smooth running system.

obtain interconnection of their offices without any change in infrastructure.

Read more about it

Wireless Communication and Networks 2nd Edition William Stallings IEEE 802.11 WLAN Specification http://grouper.ieee.org/groups/802/11/ 3rd Generation Partnership Project 2 Wireless IP Network Standard 3GPP2 P.S0001-B - Version 1.0.0 Version Date: October 25, 2002 Wireless LAN Access Network Architecture for Mobile Operators Juha Ala-Laurila et. al. , Nokia Mobile Phones, IEEE Communications Magazine, November 2001 WLAN as a Complement to GPRS and 3G Services - Dr Mario Davoli Ericsson Australia, White Paper Doc. Reference No: EPA/B 2223 158, 27 June 2002

About the authors Conclusion

We have proposed a new solution for integration of 3G and Wireless LAN. This architecture and recommendations have not been tested in the real world. But, with the available network components and standards available, a feasibility test was performed to ensure the implementation. (Ericsson is working towards a similar implementation, but using all together different technologies.) With this method, we achieved a reduction in the cost of deployment and scalability, and the employee was always connected with the same identification and standardized technology. Also, this system refrains from legal issues, since it uses unlicensed frequency band. Hence, corporations can deploy this system and Azim A. Samjani received his Bachelor of Engineering (B.E.) in Computer Engineering from University of Pune, India in August 2001 and his Master of Science in Engineering (M.S.E.) in Telecommunications and Networking from the University of Pennsylvania in May 2004. Contact: azim@computer.org Rahul R. Shetty, Contact: shettre@ hotmail.com Rohit George Mathew, Contact: rohitgmathew@yahoo.com

Acronym glossary
3G AP BS BSS CN CO 3rd Generation Wireless Systems Access Point Base Station Basic Service Set Corporate Network Central Office Distribution System Mobile Terminal Switching Office Private Branch eXchange Public Switched Telephone Network Transmission Control Protocol/ Internet Protocol Universal Serial Bus Voice over Internet Protocol (IP) Wireless Local Area Network (802.11) Wireless Private Branch eXchange

Voice / IP Packets

DS MTSO PBX

Authentication

PSTN Data Base TCP/IP VOIP Packets USB VOIP WLAN

Data

Voice IP Packet

To / From 3G LAN Server Fig. 4

WPBX

OCTOBER/NOVEMBER 2004

13