Enhance Packet Microwave Performance Through Fragmentation and FER Improvements
Packet microwave communications has rapidly gained industry momentum since the launch of the Alcatel-Lucent 9500 microwave packet radio (MPR) in February 2008. Packet microwave communications actually represents the only market segment that grew in 2011 with a compound annual growth rate (CAGR) of around 30%. All other communications segments declined in that calendar year (see, for example, refs. 1 and 2). Forecasts anticipate that packet microwave communications will grow to be the dominant microwave communications technology in a two- to three-year timeframe.
The 3GPP Long Term Evolution (LTE) network architecture has mandated the use of Internet Protocol (IP)/Ethernet packet technologies to link radio-access-network (RAN) infrastructure elements together (ref. 3). The adoption of packet technologies in microwave communications systems implies a change in the way that a microwave system behaves and complies to a specified service-level agreement (SLA), as performance and availability are now expressed through parameters relevant in the data domain.
This article explains how packet microwave technology can provide time-division-multiplex (TDM) services—with performance that is comparable, or better to, former TDM microwave systems—by means of “data awareness.” Data awareness consists of various system mechanisms including achieving improvements in fragmentation and frame error rate (FER) in order to improve the performance of a packet microwave network. Both mechanisms have been patented by Alcatel-Lucent4,5 and are implemented into the 9500 microwave-packet-radio (MPR) product family.
Defining A Packet Microwave System
The Metro Ethernet Forum (MEF) has provided a definition of a packet microwave system (ref. 6), which characterizes it as a system where a single Ethernet switching function handles all traffic. This is in contrast to hybrid microwave systems, where independent TDM and Ethernet switching functions are used to carry traditional TDM and packet backhaul, respectively (i.e., a TDM switching matrix is used to cross-connect TDM flows while an Ethernet switching function is used to forward Ethernet frames). A packet microwave system supports legacy E1/T1, ATM, and STM/SONET traffic, using technology to adapt these types of traffic into a sequence of ad-hoc Ethernet frames. In a packet microwave system, conversion of legacy traffic into Ethernet frames is performed by a dedicated interworking function (D_IWF). That function is designed to work in conjunction with the radio transmission complex in order to deal with the peculiarity of such a transmission over a radio channel.
Based on this definition, a packet microwave system is obviously more than a simple, Ethernet-based microwave communications network coupled with a generic interworking function. The D_IWF is a fundamental differentiator of packet microwave systems compared to standard Ethernet-only microwave communications systems. D_IWF characteristics are critical to the effective operation of a packet microwave communications system. After all, a packet microwave system must adopt the typical mechanisms to differentiate and handle packet services (e.g. QOS). In addition, the D_IWF makes it possible to provide TDM services with the same performance as a pure TDM radio. The bundle of these “data awareness” mechanisms, which include both fragmentation and FER improvement, is critical for guaranteeing TDM-comparable performance with a packetized system.
Model Of A Packet Radio
When a packet microwave is used to carry legacy services (e.g., TDM voice), the radio channel can be modeled as shown in Fig. 1, which is derived from ref. 4.
1. This diagram portrays a model for a packet radio and how it connects to a network.
Ethernet frames enter the system from a native Ethernet port at the first packet microwave radio. They reach that radio’s internal switch before being sent to the transmission complex, where they are modulated over the radio channel. Legacy services are converted to Ethernet form at the D_IWF. All packet traffic is sent to the packet aggregation network connected to the second packet microwave radio. Note that the inverse D_IWF is depicted in grey on the second microwave system, as it may not be necessary to apply a conversion back to TDM at this point in the network. An enhanced Ethernet-interface-over-radio (E_Eth I/F) function represents the radio transmission complex, where data-awareness mechanisms optimize the transmission on the bandwidth-constrained radio channel. This function is key to packet-microwave radios.
Looking at this system from a “black-box” perspective, it is possible to ignore how many links comprise a packet transmission chain. The main concern in characterizing system performance is in achieving an end-to-end measurement from the system through packet-oriented parameters. These parameters include frame delay, frame delay variation, and frame loss.
Understanding Fragmentation Fundamentals
Figure 2 depicts the basic behavior of fragmentation.
2. This diagram shows the essential behavior of fragmentation in a packet microwave system.
Incoming signal flows are classified. In typical network scenarios, TDM services that have been adapted to packet (circuit emulation) frames go into a high-priority queue. In contrast, fragmented data frames are sent to a low-priority queue.
An output scheduler services the traffic to be sent to the radio complex. [Note that only two queues are shown in Fig. 2 for simplicity and clarity. In a system such as the 9500 MPR, advanced scheduling mechanisms supporting eight queues are used to support optimal packet quality-of-service (QoS) treatments.] It is important that the scheduler always serves the high-priority frames first. If a low-priority segment is transmitted, a high-priority frame coming slightly after is queued. It remains in the queue for the time equivalent to the transmission of one low-priority segment.
Packet fragmentation is used to make the packet wait time deterministic, allowing TDM services to be reliably transported over packet networks with the same performance of TDM. The only latency introduced by the system is the time needed by D_ IWF to convert TDM samples into a sequence of Ethernet frames in compliance with the different standards available (see, for example, refs. 7, 8, and 9).
Note that a chain of N packet microwave links requires only one D_IWF at the ingress of the microwave network and a complementary D_IWF at the egress of the network (when legacy traffic is converted back into native TDM services). Hence, latency due to the end-to-end microwave network is minimized.
Fragmentation can be applied to several classes of service, allowing a granular handling of the traffic. This capability is often referred to as “service awareness,” a key characteristic of packet microwaves. It is the capability of guaranteeing deterministic behavior in any network condition to higher-priority traffic—with a latency that is always under control.
The fragmentation process divides Ethernet frames into segments of defined length. For example, length is selected based on the air frame of the microwave system or specific application requirements. In case of packetized TDM traffic or circuit-emulation traffic carrying voice, the length of data segments can vary within a range of a few hundred bytes. In this way, there is no risk that a frame generated by D_IWF and carrying TDM traffic will be stuck in a queue (even if high priority). If it was stuck in a queue, the frame would have to wait for the completion of the physical transmission of a long data frame, thereby introducing frame delay variation.
The fragmentation process guarantees that long data frames, such as Ethernet jumbo frames, are segmented. Thus, a high-priority voice frame contending the channel is always transmitted after the fragment that it is currently handled by the scheduler. The voice frame must wait for the end of the fragment and not for the end of the entire jumbo frame. The effect is a decrease of the maximum waiting time of a voice frame in a queue and a relevant benefit for the output wander at the receiver site, as suggested by Fig. 3.
3. This approach minimizes frame delay variations.
The first advantage of fragmentation, then, is to minimize the frame delay variation of high-priority traffic. The jitter buffers at the receiving end, which are used to compensate for the delay variation, can be quite small. This translates into a reduction of the overall system latency.
The inter-arrival profile of frames carrying voice services has its effects on the performance of clock-recovery algorithms. This is in line with the mask specified by the relevant standards, as is the case of ITU-T G.823 (ref. 10). It specifies the jitter and wander requirements at PDH User-Network Interfaces (UNIs).
Finding FER Improvement
Bit error rate (BER) is a measure of a radio channel’s quality. FER provides the same measurement of radio channel quality, but expressed as errored frames instead of errored bits. It proves to be a more effective metric for measuring Ethernet radio channel quality. An Ethernet frame is considered errored if at least one bit in the frame is errored, causing a wrong frame check sequence (FCS).
Assuming a normal error distribution, the probability of there being exactly one errored bit in an Ethernet frame is (see annex F of ref. 11 for more details):
P1 = p × (1 - p) FL × 8 - 1 × (FL × 8) where p = BER and FL is the frame length.
For FL = 64 (the minimum frame length), for example, a BER of 1 × 10-6 corresponds to a FER of 5 x 10-4.
The probability that a frame contains exactly two errors is:
P2 = p2 × (1 - p) FL × 8 - 2 × (FL × 8)(FL × 8 - 1)/2
So, given that Pi is the probability of having i errors in a frame, the FER can be expressed as:
FER = ∑i Pi where 1 ≤ i ≤ N
It is possible to generalize the formula to compute exactly N errored bits in an arbitrary frame whose length is FL. Starting from two errors onward, however, the probability of errors in a frame becomes negligible for low BER values, as it is in standard operating conditions. Hence, in the subsequent analysis, a single error is considered.
The next question is if a “right” value of FL exists to minimize the probability of achieving one error in a frame. In theory, FL should be kept at a minimum. This approach is unrealistic, however, as microwave links can represent bandwidth-constrained media. Because the total overhead increases with the number of fragments, FL should be maximized. It is easy to understand that the optimal FL is a tradeoff between these two aspects: BER/FER conversion versus spectrum efficiency. For circuit-emulated voice, FL is fixed at the optimal value for all fragments. (FL may depend on many factors, such as hardware structure, code implementation, error distribution, etc.)
In the event of an errored frame, standard Ethernet behavior is to discard it. In an advanced packet microwave system, however, this behavior is modified as follows: If the fragment with error(s) is data, the frame to which that fragment belongs is discarded. If the fragment is part of an emulated TDM circuit, the fragment is maintained and the frame is rebuilt and forwarded.
This approach is consistent with the way that errors are handled in TDM and hybrid microwave systems. Errored octets are forwarded under any conditions, which keeps BER/FER low and causes just a “glitch” (or short interruption) on a telephone conversation. To provide this level of performance, the same is done by a packet microwave on the Ethernet data path.
A correction code, protecting both the header and payload of a frame carrying circuit emulation, is computed and appended to the frame itself. This capability is not described here, but it is another component of data awareness.
In addition to the optimal transport of circuit-emulated voice, two other applications receive benefits from fragmentation. The first is the transport of packet-based synchronization, such as that done by the IEEE 188 Version 2 protocol. A low packet-delay variation requirement is addressed, as high-priority 1588 frames are serviced with a deterministic delay in queues. The second is the implementation of the Radio Link Aggregation Group (R-LAG). The R-LAG algorithms that balance traffic over multiple radio channels benefit from handling fragments that are all the same length.
Packet microwave systems represent the latest generation of microwave communications platforms and the fastest-growing microwave market segment. They support the advanced capabilities required to facilitate a migration from legacy circuit-switched services to 4G/LTE packet-optimized backhaul networks. Fragmentation and FER improvement are key technologies employed by advanced packet microwave systems. They guarantee the operation of different services including voice, synchronization, and efficient LAG distribution. In doing so, they give each service better performance than other generations of microwave platforms. Future articles in this series will explore other technical aspects of the packet microwave architecture including additional details not reported here.
Acknowledgments
Daria Cattelan and Mario Giovanni Frecassetti have extensively supported the writing of this article. The authors wish to thank several colleagues at Alcatel-Lucent for their precious support and constructive discussions including: Gianluca Boiocchi, Giuseppe De Blasio, and Marzio Gerosa—all members of the System and Architectures laboratory of the Wireless Transmission product unit—and Scott Larrigan of Product Marketing.
References
1. Sky Light Research, “Microwave Market Summary & Executive Analysis 2Q12,” September 15, 2012.
2. Dell’Oro Group, “Microwave Transmission Report 2Q12: Market Summary & Vendor Information - TDM Microwave, Packet Microwave, Hybrid Microwave - Worldwide & Regions,” August 30, 2012.
3. 3GPP TS 36.300, “Evolved Universal Terrestrial Radio Access (E-UTRA) and Evolved Universal Terrestrial Radio Access Network (E-UTRAN); Overall description; Stage 2,” V8.12.0, March 2010.
4. Patent EP1936881, “Controlling wander induced by packet delay variation in a TDM over Ethernet transmission network,” http://worldwide.espacenet.com/publicationDetails/biblio?CC=EP&NR=1936881&KC=&locale=en_EP&FT=E
5. Patent EP2048827, “Avoiding BER degradation in a TDM over Ethernet transmission network,” http://worldwide.espacenet.com/publicationDetails/biblio?CC=EP&NR=2048827&KC=&locale=en_EP&FT=E
6. Metro Ethernet Forum, “Microwave Technologies for Carrier Ethernet Services,” January 2011.
7. Metro Ethernet Forum, MEF 8, “Implementation Agreement for the Emulation of PDH Circuits over Metro Ethernet Networks,” October 2004.
8. IETF RFC 4553, “Structure-Agnostic Time Division Multiplexing (TDM) over Packet (SAToP),” June 2006
9. IETF RFC 5086, “Structure-Aware Time Division Multiplexed (TDM) Circuit Emulation Service over Packet Switched Network (CESoPSN),” December 2007.
10. ITU-T G.823, “The control of jitter and wander within digital networks which are based on the 2048 kbit/s hierarchy,” March 2000.
11. ETSI EN 302 217-2-2 V2.0.0 (draft version), “ Fixed Radio Systems; Characteristics and requirements for point-to-point equipment and antennas; Part 2-2: Digital systems operating in frequency bands where frequency co-ordination is applied; Harmonized EN covering the essential requirements of article 3.2 of the R&TTE Directive,” September 2012.