Optical packet marking for fast discarding in an OLS scheme

نویسندگان

  • J. J. Vegas Olmos
  • I. Tafur Monroy
  • A. M. J. Koonen
چکیده

We propose and validate experimentally a Time-to-Live signaling system for an Optical Label Swapping scheme based on 10 Gbit/s DPSK packets and with 100 Mbit/s SCM label. The proposed scheme allows fast packet discarding by using a 3.5 GHz sub-carrier tone. DPSK payload has only a 2 dB power penalty at 10 –9 bit error-rate after superimposing the SCM and TTL labeling signal. Introduction Due the continuing growth of the Internet and the introduction of high-bit rate WDM connections in metro and backbone networks, the need for highly efficient flexible switching solutions is becoming more apparent. One solution is optical label swapping (OLS), that enables the implementation of packet routing and forwarding functions in IP-over-WDM [1]. By using short fixed-length labels the core nodes of the network forward/switch packets fast and efficiently while keeping the payload data entirely in optical domain. Previous work has been done in label encoding for DPSK signals [2-3]. However, our proposed labeling technique combines angle modulation and sub-carrier multiplexing (SCM), namely the payload data is encoded in differential phase shift keyed (DPSK) modulation format and the label information is transported by using SCM techniques. The robustness against chromatic dispersion, polarization mode dispersion (PMD), and cross-phase modulation (XPM) effects during transmission along optical fibers has been reported previously for this scheme [4], as well as record-high capacity and link reach optical transmission by using the DPSK modulation format [5-6]. OLS also offers the advantage of being protocol transparent, compatible with legacy and emerging network technologies by adopting the GMPLS frame work for a unified control plane. One problem faced by burst and packet switched networks is the “routing loops”, where mis-directed or mis-labeled packets are routed in circles without reaching their destination [7], producing network congestion. To minimize this problem, MPLS and GMPLS [8-9] incorporate a “time-to-live”(TTL) field that is decremented by one at each “hop” and when the value reaches zero, the packet is dropped from the network. We propose and demonstrate experimentally an inband signaling system by introducing the TTL value state (still valid packet or discarded packet) in the optical domain, allowing fast recognition of the packet state in the node. If the packet is to be discarded, it will be dropped even before reading the label and so saving time and node resources, compared to the broadband wavelength systems proposed previously [10]. The TTL value will remain in the corresponding label field, but it could be also introduced in the new tone. Our technique utilises the DPSK modulation to convey the payload information. An SCM signal carried at 1 GHz is used for the label data and a sinusoidal tone at 3.5 GHz for the TTL state. The label and TTL swapping is performed by using a semiconductor optical amplifier (SOA) as eraser. Experiments and Results The experimental setup is shown in Fig. 2. A distributed feed-back (DFB) laser source with an integrated Electro-absorption Modulator (EAM) is used for inserting an RF signal at 1 GHz with 100 Mbit/s amplitude modulation and the sinusoidal tone at 3.5 GHz, both of them with 100 mV RMS, onto the optical carrier operating at the 1555.37 nm wavelength. The DFB section is biased at 80 mA while the EAM section is reverse biased at 23 mA. A LiNbO3 phase modulator is used to impose DPSK modulation at 10 Gbit/s using a 2-1 PRBS pattern. Figure 1: Experimental set-up. Using gain saturation in a semiconductor optical amplifier (SOA) the Label and the TTL signal can be removed. The average output power of the phase modulator was –8.9 dBm, while no variation was observed of this value if either the label and/or the 3.5 GHz TTL signal were present. After extracting the back-to-back performance, the power of the signal was adjusted to –3.5 dBm by varying the bias current of the DFB laser to 90 mA and then it was fed into the SOA. This adjustment was done in order to find the optimal point for gain saturation in the SOA, related to the equations we developed. In Fig. 2 is shown the RF spectrum of the signal before the SOA and after the erasure process of the label and the TTL mark. Figure 2: RF spectrum of the original labeled and TTL signaled signal before the SOA (a), and after the erasure process (b). The insets show the eye diagram of the DPSK signal in each scenario. We employed a single photodetector receiver for DPSK detection. After the erasure process the eye diagram became noisy but the eye is kept open enough for error-free detection. Fig. 3 shows the bit error rate (BER) curves obtained for the DPSK unlabeled and after the SOA (including with and without the label and the TTL signal). Figure 3: BER curves of the labeled and unlabeled DPSK signal, and the DPSK signal after the erasure process. The DPSK unlabeled receiver sensitivity was measured to be –20.3 dBm for a BER<10, and the combined effect of superimposing the label and the TTL signal introduces 2 dB penalty in total. For the DPSK signal after the erasure process, the received optical power level yielding a BER of 10 was measured to be –19.2 dBm. Therefore, only 1.1 dB penalty was suffered in the SCM and TTL erasure process. This power penalty might be reduced by using a balanced receiver configuration and by proper optimization of the modulation index for the SCM and TTL signals to trade-off SCM and DPSK performances. In OLS networks, core nodes perform the function of label reading and erasure that we demonstrate in this paper. Label insertion and TTL signaling can be reinserted by using an EAM, similarly to the one used in our experiment for label generation. Therefore, this architecture offers the prospect of integration on a photonic circuit. ConclusionsWe have presented an optical labeling switchingtechnique based on the use of DPSK modulation forthe high-bitrate payload data, SCM for the labelinformation, and sinusoidal tone for the TTL signaling.We have experimentally demonstrated its feasibilityfor 10 Gbit/s DPSK, 100 Mbit/s AM-SCM at 1 GHzlabel data and presence/no presence of tone at 3.5GHz. We have experimentally demonstrated the useof an SOA as low band pass filter. The power penaltydue to insertion of superimposed SCM and TTLsignals is 2.0 dB. After the SCM and TTT erasureprocess a power penalty of 1.1 dB was measured,compared to the back-to-back DPSK performance. References1 C. Qiao et al., OFC, 1(2003), 219-2202 N. Chi et al., Electronics Letters, 39(2003).3 X. Liu et al., ECOC 2003, Tu4.4.3.4 M. Rohde et al., Electronics Letters, 36(200), 1483.5 B. Zhu et al., Journal of Lightwave Technology, 22(2004).6 C. Rasmussen et al., Journal of LightwaveTechnology, 22 (2004).7 V. Paxson, IEEE/ACM Trans. Networking, 5(1997), 601-615.8 L. Berger, RFC memo 3471 (2003).9 P. Agarwal et al., RFC memo 3443 (2003).

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

VRED: An improvement over RED algorithm by using queue length growth velocity

Active Queue Management (AQM) plays an important role in the Internet congestion control. It tries to enhance congestion control, and to achieve tradeoff between bottleneck utilization and delay. Random Early Detection (RED) is the most popular active queue management algorithm that has been implemented in the in Internet routers and is trying to supply low delay and low packet loss. RED al...

متن کامل

VRED: An improvement over RED algorithm by using queue length growth velocity

Active Queue Management (AQM) plays an important role in the Internet congestion control. It tries to enhance congestion control, and to achieve tradeoff between bottleneck utilization and delay. Random Early Detection (RED) is the most popular active queue management algorithm that has been implemented in the in Internet routers and is trying to supply low delay and low packet loss. RED al...

متن کامل

An adaptive probabilistic marking scheme for fast and secure traceback

IP traceback can be used to find direct generator(s) and path(s) of attacking traffic. Probabilistic marking schemes, as one type of IP traceback technologies, have been most studied, but they are difficult to fast reconstruct attacking path(s) and defend against spoofed marks generated by attacking source(s). In this paper, we present Adaptive Probabilistic Marking scheme (APM). In APM, when e...

متن کامل

Frame-Induced Packet Discarding: An Efficient Strategy for Video Networking

In order to provide efficient frame loss guarantees for video communication over ATM-like fast packet switched networks, we propose a simple to implement, yet effective, strategy called Frame-Induced Packet Discarding (FIPD), in which, upon detection of loss of a threshold number (determined by an application’s video encoding scheme) of packets belonging to a video frame, the network attempts t...

متن کامل

An Efficient Packet Discarding Scheme for On-demand Video Streaming

Streaming services over the Internet are expected to grow rapidly in the near future. However, bursty packet losses caused by buffer overflow at routers are a serious problem in these services, and are responsible for degradation of the Quality of Service (QoS). Active Queue Management (AQM) that discards the packets at routers is effective for avoiding the buffer overflow but results in the de...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2004