LAN implementation | Microwave

LAN implementation

There are some short haul microwave systems used for corporate LANs that do require an FCC license because they operate in a more crowded portion of the spectrum. This is the case with the 18-GHz frequency band, which is used by Motorola's Altair product line, for example. The FCC allocates licenses for the 18-GHz region of the spectrum for five channels, each consisting of two 10-MHz bands in the 18- to 19-GHz band. The two-channels-per-license arrangement permits full-duplex communications using one channel in each direction.

While a license may be construed as a liability, it offers the guarantee of interference-free communication. Unlike unlicensed spectrum, such as that often used by spread-spectrum technologies, licensed spectrum gives the license-holder a legal right to an interference-free data communications channel. Owners of wireless LANs operating on unlicensed spectrum are continuously at risk of unauthorized interference destroying their data communications capabilities, and do not always have legal recourse.

Since Motorola has obtained licenses for 18-GHz operation in all U.S. metropolitan areas with populations above 30,000, Altair microcells in these areas are fairly well protected from unauthorized interference. Motorola customers do not have to deal directly with the government or wait for approval to operate Altair equipment. Upon purchasing the equipment, they fill out a simple registration form, and immediately place their equipment into operation. Motorola's Frequency Management Center dedicates and coordinates frequencies to specific customer locations. In the event a customer moves the Altair system to another location, Motorola provides an 800 number to report that fact in compliance with FCC requirements. Motorola handles frequency coordination for the customer at the new location.

LAN extension

Corporations are making greater use of wireless technologies for extending the reach of LANs where a wired infrastructure is absent or costly. Wireless bridges and routers can extend data communication between buildings in a campus environment or between buildings in a metropolitan area.

Wireless bridges. Short-haul microwave bridges equipped with directional antennas provide an economical alternative to leased lines or underground cabling. Because they operate over very short distances—less than a mile—and are less crowded, they are less stringently regulated and have the additional advantage of not requiring an FCC license.

The range of the bridge is determined by the type of directional antenna. A 4-element antenna, for example, provides a wireless connection of up to 1 mile. A 10-element antenna provides a wireless connection of up to 3 miles.

Directional antennas require a clear line of sight. To ensure accurate alignment of the directional antennas at each end, menu-driven diagnostic software is used. Once the antennas are aligned, and the system ID and channel are selected with the aid of configuration software, the system is operational. Front-panel LEDs provide a visual indication of link status and traffic activity. The bridge unit has a diagnostic port, allowing performance monitoring and troubleshooting through a locally attached terminal or remote computer connected via modem.

Because it is fully compatible with the IEEE 802.3 Ethernet standard, microwave bridges support all Ethernet functionality and applications without the need for any special software or network configuration changes. For Ethernet connections, the interface between the microwave equipment and the network is virtually identical to that between the LAN and any cable medium, where retiming devices and transceivers at each end of the cable combine to extend the Ethernet cable segments. Typically, microwave bridges support all Ethernet media types via AUI connectors for thick Ethernet (10Base5), 10Base2 connectors for thin Ethernet, and twisted-pair connectors for 10BaseT Ethernet. These connections allow microwave bridges to function as an access point to wired LANs.

Like conventional Ethernet bridges, microwave bridges perform packet forwarding and filtering to reduce the amount of traffic over the wireless segment. The microwave bridges contain Ethernet address filter tables, which help to reduce the level of traffic through the system by passing only the Ethernet packets bound for an interbuilding or intrabuilding destination over the wireless link. Since the bridges are "self-learning," the filter tables are automatically filled with Ethernet addresses as the bridge learns which devices reside on its side of the link. That way, Ethernet packets that are not destined for a remote address remain local. The table is dynamically updated to account for equipment that is either added or deleted from the network. The size of the filter table can be 1000 entries or more, depending on vendor.

With additional hardware, microwave bridges have the added advantage of pulling double-duty as a backup to local T1/E1 facilities. When a facility degrades to a pre-established error-rate threshold or is knocked out of service entirely, the traffic can be switched over to a wireless link to avoid loss of data. When line quality improves or the facility is restored to service, the traffic is switched from the wireless link to the wireline link.

Wireless routers. Wireless remote access routers scale wireless connect geographically disbursed LANs by creating a wireless WAN over which network traffic is routed at distances of 30 miles or more using spread-spectrum technology.

Unlike wireless bridges which simply connect LAN segments into a single logical network, wireless routers function at the network layer with IP/IPX routing, permitting the network designer to build large, high performance, manageable networks. Wireless routers are capable of supporting star, mesh, and point-to-point topologies that are implemented with efficient MAC protocols. These topologies can even be combined in an internetwork.

A polled protocol (star topology) provides efficient shared access to the channel even under heavy loading (Figure 1). For small-scale networks, a CSMA/CA protocol supports a mesh topology (Figure 2). Clusters of nodes can be connected using a point-to-point protocol when building large-scale internetworks (Figure 3). The single-hop node-to-node range is up to 30 miles, depending on such factors as terrain and antennas, with a multiple hop range extending on the order of a hundred miles.

Figure 1: In the star topology, remote stations interconnect with the central base station, and with other remote stations, through the base station. Only one location needs line-of-sight to the remotes. Networks and workstations at each location tie into a common internetwork. The maximum range between the central base station and remote stations is approximately 15 miles.

Figure 2: In the mesh topology, each site must be line-of-sight to every other. A CSMA/CA protocol ensures efficient sharing of the radio channel. The range with omni-antennas is up to 3.5 miles.

Figure 3: The point-to-point topology is useful where there are only two sites. It can also function as a repeater link between clusters of sites. The range using directional antennas is up to 30 miles.

End-to-end SNMP supports management of the radios and the wireless WAN along with the remainder of the enterprise network using industry standard SNMP tools. The result is a manageable network with reach extending to metropolitan, suburban, rural, remote, and isolated areas.

Applications of wireless routers include remote site LAN connectivity and network service dissemination. Organizations with remote offices such as banks, health care networks, government agencies, schools, and other service organizations can connect their computing resources. Industrial and manufacturing companies can reliably and cost-effectively connect factories, warehouses, and research facilities. Network service providers can distribute Internet, VSAT, and other network services to their customers.

A WAN built with wireless routers exploits the tariff-free wireless "infrastructure." A wireless WAN offers a substitute for a wired infrastructure with its associated costly service fees. In areas where a wired infrastructure is absent or underdeveloped, the use of wireless routers newly enables internetworking. Performance is comparable to commonly used wired WAN connections, approaching T1 speeds with a 1.3 Mbps data rate.

A user-supplied PC hosts the wireless router device through a connection to the PC's parallel printer port. Software includes an installation utility, network drivers, management, and IP/IPX routing. No radio license is required in the United States and many other countries, simplifying the network development process compared to licensed microwave.

Frequency hopping | Spread-Spectrum LANs

Frequency hopping entails the transmitter jumping from one frequency to the next at a specific hopping rate in accordance with a pseudo-random code sequence. The order of frequencies selected by the transmitter is taken from a predetermined set as dictated by the code sequence. For example, the transmitter may have a hopping pattern of going from channel 6 to channel 1 to channel 5 to channel 9 to channel 3 and so on. The receiver tracks these changes. Since only the intended receiver is aware of the transmitter's hopping pattern, then only that receiver can make sense of the data being transmitted.

Other frequency hopping transmitters will be using different hopping patterns that usually will be on noninterfering frequencies. Should different transmitters coincidentally attempt to use the same frequency and the data of one or both become garbled at that point, retransmission of the affected data packets is required. Those data packets will be sent again on the next hopping frequency of each transmitter. Most LAN protocols have an integral error detection capability. When the protocol's error checking mechanism recognizes incoming packets that are bad or determines that there are missing packets, the receiving station requests a retransmission of only those packets. When the new packets arrive to rendezvous with those held in queue, the protocol's sequencing capability puts them in the correct order.

Some wireless LAN vendors use algorithms for dynamic data rate switching in conjunction with frequency hopping spread spectrum. The algorithms dynamically determine without user intervention whether the wireless signal is uncorrupted and whether the number of retransmissions is sufficiently low to allow data to be transmitted at a higher speed. The transmitting device can then select the maximum reliable data rate on the fly.

As noted, the FCC has authorized the use of spread-spectrum transmitters without requiring individual user licenses. However, only direct sequence or frequency hopping techniques may be used—no other spreading techniques are permitted, per FCC Rule 15.247. Users may define their own channel width up to a maximum of 500 kHz in the 900-MHz band or 1 MHz in the 2.4-GHz band. Frequency hopped systems must not spend more than 0.4 seconds on any one channel each 20 seconds, or 30 seconds in the 2.4-GHz band. Furthermore, they must hop through at least 50 channels in the 900-MHz band, or 75 channels in the 2.4-GHz band. These rules reduce the chance of repeated packet collisions in areas with multiple transmitters.

Direct sequence spread spectrum offers better performance, but frequency-hopping spread spectrum is more resistant to interference and is preferable in environments where many other devices generate electromechanical noise. Direct sequence is more expensive than frequency hopping and uses more power.

What hardware the user chooses will largely be dictated by applications. Users concerned with good performance in environments where interference is not a problem, will generally opt for the direct sequence solution. Users who need a small, inexpensive portable wireless adapter for their notebook or PDA, generally go for frequency-hopping spread spectrum.

Spread-spectrum technology is also used in bridges and routers for LAN extension. Although line-of-sight connections are required, such devices allow users to extend the reach of wireless LANs (Ethernet and token ring) by as much as 25 miles. For example, a bridge can be used to connect to a central server to access e-mail, groupware, and client/server applications without any modifications. Bridges and routers can also be used to link wireless and wireline LANs together into a seamless enterprise network.

Direct sequence | Spread-Spectrum LANs

In direct sequence spreading, the radio energy is spread across a larger portion of the band than is actually necessary for the data. This is done by breaking each data bit into multiple sub-bits called "chips" to create a higher modulation rate. The higher modulation rate is achieved by multiplying the digital signal with a chip sequence. If the chip sequence is ten, for example, and it is applied to a signal carrying data at 300 Kbps, then the resulting bandwidth will be ten times wider. The amount of spreading is dependent upon the ratio of chips to each bit of information. Because data modulation widens the radio carrier to increasingly larger bandwidths as the data rate increases, this chip rate of 10 times the data rate spreads the radio carrier to 10 times wider than it would otherwise be for data alone. Figure 8.2 compares the bandwidth required for 300 Kbps of data and the 10:1 increase in bandwidth due to spreading.

Figure 1: Spreading 300 Kbps of data across a wider band causes it to resemble random noise during transmission, making it difficult to intercept. (a) Original signal. (b) Spread signal.

The rationale behind this technique is that a spread-spectrum signal with a unique spread code cannot create the exact spectral characteristics as another spread-coded signal. Using the same code as the transmitter, the receiver can correlate and collapse the spread signal back down to its original form, while other receivers using different codes cannot.

This feature of spread spectrum makes it possible to build and operate multiple networks in the same location. By assigning each one its own unique spreading code, all transmissions can use the same frequency band, yet remain independent of each other. The transmissions of one network appear to the other as random noise and are filtered out because the spreading codes do not match.

This spreading technique would appear to result in a weaker signal-to-noise ratio, since the spreading process lowers the signal power at any one frequency. Normally, a low signal-to-noise ratio would result in damaged data packets that would require retransmission. However, the processing gain of the despreading correlator recovers the loss in power when the signal is collapsed back down to the original data bandwidth. This process is not the same as the signal enhancement techniques used by certain devices on wireline networks, since the signal is not strengthened beyond what would have been received had the signal not been spread. (In wireline data transmission, DSUs, for example, perform data regeneration by reshaping the transmit signal before it is sent out over the digital facility to ensure optimal network performance.)

The FCC has set rules for direct sequence transmitters. Each signal must have ten or more chips. This rule limits the practical raw data throughput of transmitters to 2 Mbps in the 902-MHz band and 8 Mbps in the 2.4-GHz band. Unfortunately, the number of chips is directly related to a signal's immunity to interference. In an area with a lot of radio interference, users will have to give up throughput to successfully limit interference.

Fundamental LAN Planning Guidelines

The Cisco guide recommends a detailed analysis of the following LAN elements:

• LAN/campus topology

• IP addressing plan

• Location of TFTP servers, DNS servers, DHCP servers, firewalls, network address translation (NAT) gateways, and port address translation (PAT) gateways

• Potential location of gateways and call telephony servers

• Protocol implementation including IP routing, Spanning Tree, VTP, IPX, and IBM protocols

• Device analysis including software versions, modules, ports, speeds, and interfaces

• Phone connection methodology (direct or daisy chain)

According to the Cisco guide, the significant LAN topology issues are:

• Available average bandwidth

• Available peak or burst bandwidth

• Resource issues can may affect performance including buffers, memory, CPU, and queue depth

• Network availability

• IP phone port availability

• Desktop/phone QoS between user and switch

• Network scalability with increased traffic, IP subnets, and features

• Back-up power capability

• LAN QoS functionality

• Convergence at Layers 2 and 3

IP addressing issues that should be reviewed are:

• Phone IP addressing plan

• Average user IP subnet size use for the campus

• Number of core routes

• IP route summary plan

• DHCP server plan (fixed and variable addressing)

• DNS naming conventions

Potential considerations with IP addressing include:

• Route scalability with IP phones

• IP subnet space allocation for phones

• DHCP functionality with secondary addressing

• IP subnet overlap

• Duplicate IP addressing

The locations (or potential locations) of servers and gateways are important to ensure that service availability is consistent across the LAN infrastructure and for multiple sites. Gateways and servers in the review should include:

• TFTP servers

• DNS servers

• DHCP servers

• Firewalls

• NAT or PAT gateways

• Call telephony server

• Gateway location

After determining the location of these network elements, the following issues should be analyzed:

• Network service availability

• Gateway support (in conjunction with the IP telephony solution)

• Available bandwidth and scalability

• Service diversity

IP telephony scalability and availability issues will be affected by protocols in the network. The following areas for the protocol implementation analysis are:

• IP routing including protocols, summarization methods, non-broadcast media access (NBMA) configurations, and routing protocol safeguards

• Spanning Tree configuration including domain sizes, root designation, uplink fast, backbone fast, and priorities in relation to default gateways

• HSRP configuration

• VTP and VLAN configuration

• IPX, DLSW, or other required protocol services, including configuration and resource usage

With regard to protocol implementation, the following issues should be reviewed:

• Protocol scalability

• Network availability

• Potential impact on IP telephony performance or availability

All network devices should be reviewed and analyzed to determine whether the network has the desired control plane resources, interface bandwidth, QoS functionality, and power management capabilities. The checklist for this process includes:

• Device (type and product ID)

• Software version(s)

• Quantity deployed

• Modules and redundancy

• Services configured

• User media and bandwidth

• Uplink media and bandwidth

• Switched versus shared media

• Users per uplink and uplink load sharing/redundancy

• Number of VLAN supported

• Subnet size, and devices per subnet

For establishing a network baseline, it is important that the following measurements be made to determine voice quality levels and potential problem areas:

• Device average and peak CPU

• Device average and peak memory

• Peak backplane use

• Average link use (prefer peak hour average for capacity planning)

• Peak link use (prefer 5 minute average or smaller interval)

• Peak queue depth

• Buffer failures

• Average and peak voice call response times (before IP telephony implementation)

Cabling questions that may help determine the readiness of the infrastructure for IP telephony readiness include:

• Does the building wiring conform to EIA/TIA-568A?

• Does your organization comply with National Electric Code for powering and grounding sensitive equipment?

• Does your organization comply with the more rigorous IEEE 1100-1992 standard for recommended practices of grounding and powering sensitive equipment?

• Does the organization have standards for data center and wiring closet power that include circuit distribution, available power validation, redundant power supply circuit diversity, and circuit identification?

• Does the organization use UPS and/or generator power in the data center, wiring closet, phone systems, and internetworking devices?

• Does the organization have processes to SNMP manage or periodically validate and test back-up power?

• Does your business experience frequent lightning strikes? Are there other potential natural disasters?

• Is the wiring to your building above ground?

• Is the wiring in your building above ground?

Network bandwidth consumption is required for each VoIP stream. In any conversation, two such streams are required: one in each direction. The required bandwidth per conversation will be based on several factors, but of primary importance is the codec used to digitize, compress, and convert an analog voice signal into IP format. The two codecs of most interest are G.711 and G.729A. G.711 is the TIA recommended codec to optimize IP telephony QoS because it reduces impairment of the voice signal across the network, but the signal is uncompressed and requires a high amount of bandwidth. To save on network transmission costs, G.729A is used for off-premises traffic because it uses a compression algorithm to reduce bandwidth requirements.

Network performance and capacity planning help to ensure that the network will consistently have available bandwidth for data and VoIP traffic and that the VoIP packets will consistently meet delay and jitter requirements. Cisco recommends the following six-step process for network capacity and performance planning:

1. Determine baseline existing network use and peak load requirements

2. Determine VoIP traffic overhead in required sections of the network based on busy hour estimates, gateway capacities, and/or CallManager capacities

3. Determine minimum bandwidth requirements

4. Determine the required design changes and QoS requirements based on IP telephony design recommendations and bandwidth requirements (overprovide where possible)

5. Validate baseline performance

6. Determine trunking capacity

LAN/WAN Design Guidelines for VoIP

There is no standard definition for QoS as it applies to real-time voice communications carried over an Ethernet LAN or IP WAN. As applied to a circuit switched PBX, QoS means consistent, reliable service delivery of control and communications signals in support of customer needs. This definition also can be used for LAN QoS in support of IP telephony. To enable LAN QoS requires all network elements, at all network layers, to work together to support a required level of traffic and service.

An IP-PBX by definition is not a virtual circuit switched communications system, like a traditional PBX, but rather a system that uses an IP network infrastructure. An IP network makes more efficient use of available bandwidth resources than does a circuit switched PBX and is designed to support the “bursty” nature of data communications traffic rather than the continuous traffic flow of real-time voice communications. IP networks can adapt to changing traffic conditions, but the level of service can be unpredictable. When used to support an IP-PBX system, the IP network must be properly designed and engineered to support the unique real-time traffic requirements of voice as opposed to less stringent data communications requirements.

QoS techniques manage bandwidth according to different application demands and network management settings but cannot guarantee a service level if resources are not available and allocated. Reserving resources for voice communications can seriously affect other network traffic. A priority for QoS network designers has been to ensure that best-effort traffic is available after resource allocations have been made. QoS-enabled high-priority voice applications must not harm lower-priority data applications.

The Internet was based on a dumb network concept with intelligent endpoints to transmit and receive datagram packets flowing through a series of network routers. IP does not deliver reliable service over the Internet: packets can be dropped by routers and are retransmitted as necessary. The service mechanism can assure data delivery, but not timely delivery. This “best-effort” service may be adequate for data networking services, but it is not good enough for voice communications.

Audio and video traffic demands sufficient bandwidth with low-latency requirements when used in two-way communications. A major challenge for network planners is to design a LAN infrastructure that satisfies an acceptable QoS level that PBX system users have grown accustomed to for their voice communications applications. A newly installed IP-PBX system in a green field location provides an ideal situation, but if a network is already installed and operating, introducing IP telephony-grade QoS should not disrupt existing services and applications.

LAN QoS levels fluctuate over time due to unanticipated changes in customer usage patterns and traffic flow. If QoS is degraded for short periods, it may significantly affect IP telephony services in ways noticeable by all system users, even if data communications services appear satisfactory. There are several reasons QoS can change:

• Temporary excessive network usage

• Insufficient link capacity

• Insufficient switch/router resources

• Traffic flow peaks

• Traffic flow interference

• Improper use of resources

Several basic control methods can be employed to manage QoS levels to ensure the higher grade of service level required by real-time voice communications:

• Reserving fixed bandwidth for mission-critical voice communications applications

• Restricting network access and usage for defined users or user groups

• Assigning traffic priorities

• Designating which kinds of traffic can be dropped when congestion occurs

There are several high-level decisions facing network planners and managers regarding the type of QoS-based network to be designed and operated. The network planner must decide whether network users are involved in the QoS functions or whether the network is in total control of QoS functions. If a network user has knowledge of QoS functions and a limited degree of QoS control, the network QoS is said to be implicit. If network QoS functions are predetermined and only the network administrator can program changes when needed, the network QoS is said to be explicit.

Another planning issue is whether QoS is soft or hard. Network QoS is said to be soft when there is no formal guarantee that target service levels will be met, even if QoS functions are implemented. Hard network QoS is a guarantee of service at a predefined level of QoS. Hard network QoS is usually available only with connection-mode transport, such as ATM constant bit rate (CBR) service.

Network QoS is also manageable by network design, by installing the necessary physical resources to support target service levels. IP-PBX system voice quality and availability can be determined by the physical LAN infrastructure and available cable bandwidth. Cisco Systems, a leading IP-PBX system supplier and the dominant supplier of data communications systems, has developed and published an IP telephony network planning guide. The Cisco guide is a planning tool for its CallManager IP-PBX system customers, but it is also useful as a network design guide for customers who plan to install and operate any converged or client/server IP-PBX system.

Dispersed Common Equipment over LAN/WAN Infrastructure

Support of IP endpoints, stations, and/or trunks may not be the only trademark of a converged IP-PBX system design. A PBX system that supports neither IP stations nor trunks can be considered an IP-PBX if the system design includes geographically dispersed port cabinets/carriers using an IP LAN/WAN infrastructure for control signaling from the common control complex and voice communications between dispersed port interface equipment. Using a LAN/WAN infrastructure to support customer communications requirements across one or more customer premises locations based on a single IP-PBX system offers several potential key benefits:

• Single system image (numbering plan, features, systems management)

• Reduced networking costs between customer locations

• Scalable system expansion

There are several types of converged IP-PBX system designs in this category. They include upgrades of traditional circuit switched PBXs and IP-PBX systems whose original design was based on a LAN/WAN infrastructure to support desktop and networking communications requirements. The latter system designs were introduced by nontraditional PBX suppliers that entered the market during the late 1990s.