This paper presents different strategies to define the architecture of a Radio-Over-Fiber (RoF) Access networks enabling Peer-to-Peer (P2P) functionalities. The architectures fully exploit the flexibility of a wavelength router based on the feedback configuration of an Arrayed Waveguide Grating (AWG) and an optical switch to broadcast P2P services among diverse infrastructures featuring dynamic channel allocation and enabling an optical platform for 3G and beyond wireless backhaul requirements. The first architecture incorporates a tunable laser to generate a dedicated wavelength for P2P purposes and the second architecture takes advantage of reused wavelengths to enable the P2P connectivity among Optical Network Units (ONUs) or Base Stations (BS). While these two approaches allow the P2P connectivity in a one at a time basis (1:1), the third architecture enables the broadcasting of P2P sessions among different ONUs or BSs at the same time (1:M). Experimental assessment of the proposed architecture shows approximately 0.6% Error Vector Magnitude (EVM) degradation for wireless services and 1 dB penalty in average for 1x10−12 Bit Error Rate (BER) for wired baseband services.
©2010 Optical Society of America
Nowadays P2P applications and video streaming occupy a significant amount of Internet bandwidth. BitTorrent and eDonkey applications have been some of the most common P2P systems and the trend continuously grows with an expected increment of video traffic for mobile users. Most of current P2P traffic is video, usually file downloads, but in the future it is expected that P2P may take a more fundamental role in downloading non-real-time Video-on-Demand (VOD) and delivering Internet video and Internet Protocol Television (IPTV) which day by day are becoming more popular [1,2]. However, due to the particular characteristics of P2P traffic, in many cases, P2P sessions go across the core network and perhaps through multiple Internet Service Providers (ISPs) even though the content must be delivered into the coverage of the Optical Line Termination (OLT)/Central Station (CS) or in a location close to the source. Therefore, it is expected that future access networks offer the functionality to transparently resolve and manage the P2P sessions for both fixed and mobile users in the access domain in order to decongest the core network and maximize the capacity with the eventual result of the Quality-of-Service (QoS) improvement. Moreover, we found that P2P connectivity can also be exploited to solve the issues with the wireless backhaul for the 4th generation of cellular wireless standards (4G). The standard, as defined by the 3rd Generation Partnership Project (3GPP), requires full mesh connectivity among (BS) due to the flat all-IP architecture and direct connectivity between the Central Station (CS) and (BS) without any radio network controller. Hence, the BSs carry more intelligence for functionalities like radio resource management towards a distributed decision system. This fact requires inter-BS logical connectivity that existing backhaul networks do not naturally support . Figure 1 shows the environment for such convergent scenario where ONUs and BS share the same optical platform for dynamic channel distribution and physical P2P connectivity.
From the network point of view, cross-layer interactions have to be considered in order to provide a broad class of services to heterogeneous users at high data rates. In particular, due to the fact that concurrent transmissions interfere among each other appropriate scheduling algorithms need to be introduced to mitigate the effects of unwanted interference. Although scheduling is a decision of the Medium Access Control (MAC) layer, the ability to schedule two or more users concurrently depends on the bit rates and transmission power as well as on the underlying channel conditions. Thus, there is an inherent coupling between the MAC and the physical layer. Taking the cross-layer issues one step further, the routing algorithms also need to take into account information regarding the physical layer as well as the scheduling decisions at any given time in order to send the packets through the routes among diverse peers using links with the highest capacities.
This paper presents different strategies for a RoF access network that enable P2P physical connectivity in a converged wired/wireless access network. Unlike the approach described in , the proposed architectures allow the simultaneous P2P and uplink connectivity for different ONU and BS, hereafter ONU/BS, while the mesh-like connectivity can be exploited to solve the issues with the 4G backhauling. To the best of our knowledge these architectures constitute a first approach to implement a convergent wired/beyond 3G wireless access network.
The paper is organized as follows: Section 2 describes the three proposed architectures along with the experimental environment used to assess their performance. Section 3 presents the performance evaluation for each one of the approaches and finally section 4 summarizes the paper.
2. System description
This section describes the proposed architectures for P2P physical connectivity in RoF access networks conveying converged wired and wireless signals. The P2P connectivity is offered by the proposed wavelength router based remote node that is able to route transparently signals among different ONU/BS.
The architecture is based on an Arrayed Waveguide Grating (AWG) and a Micro-Electro-Mechanical (MEM) based optical switch in feedback configuration. A comprehensive description of the underlying architecture can be found in . The wavelength router assigns a fixed wavelength channel and one or more extra capacity channels to each ONU/BS depending on the demand. Four fixed channels (λ1 = 1546.65 nm, λ2 = 1547.4 nm, λ3 = 1548.2 nm and λ4 = 1549.04 nm) and two extra-capacity wavelength channels (λ5 = 1549.8 nm and λ6 = 1550.6 nm) were used to experimentally demonstrate the operation among four ONU/BS. The fixed channels convey a composed Subcarrier Multiplexing (SCM) signal with baseband data at 2.5 Gb/s and a RF subcarrier with 10 Mbauds QPSK onto 5 GHz. One of the extra capacity channels transports baseband data at 2.5 Gb/s and the other carries a RF subcarrier with 10 Mbauds QPSK at 2.5 GHz. Both P2P and upstream signals transport combined baseband at 2.5 Gb/s and a RF subcarrier with 5 Mbauds QPSK onto 2.5 GHz. Baseband services are NRZ encoded with 231-1 Pseudorandom bit sequence as detailed in Table 1 . The optical downstream channels were equalized by means of a variable attenuator in the central office and system losses were compensated by an Erbium Doped Fiber Amplifier (EDFA) after 20 km transmission and prior to be launched into the dynamic router. The description of the P2P approaches is presented in the following subsections.
2.1 Time Division Multiplexing (TDM) architecture with dedicated wavelength (1:1)
The first architecture is based on the dedicated use of a wavelength to distribute the P2P services enabling 1:1 connectivity between two ONU/BSs. It means that P2P sessions among different ONU/BSs rely on TDM techniques as the P2P wavelength is transmitted to each ONU/BS in different time slots. The architecture is shown in Fig. 2(a) . A dedicated tunable laser in the remote node transmits the unmodulated optical carrier (P2P wavelength) to the ONU/BS. In this approach, the P2P channels use the next upper spectral period of the AWG respect to the downstream channels, so that according to the AWG periodical response, a given ONU/BS will be reached by its fixed downstream channel and also by the corresponding one in the upper band. In parallel, the extra capacity channels are fed backwards and switched respectively to the AWG inputs which in combination with the state of the switch assign dynamically different wavelengths at each ONU/BS. In this approach, the P2P is modulated onto the P2P wavelength and the uplink is modulated onto the same fixed wavelength that was assigned in the downlink to the ONU/BS using a wavelength rewriter based on data extinction technique by gain saturation in a Semiconductor Optical Amplifier (SOA), as seen in inset of Fig. 2(a).
Here, only as a reference the wavelength rewriter is briefly described, a comprehensive description can be found in . The saturation mode of the SOA is set by the output signal from an Erbium Doped Fiber Amplifier (EDFA) placed at the Remote Node (RN). Once the signal is launched into the network each SOA at each ONU/BS reaches the saturation regime. As a result, the difference between the low and high logic levels is considerably reduced at the SOA output so that the baseband signal modulation pattern is partly erased. Then the Fabry-Perot Filter (FPF) removes the RF subcarrier and the Mach-Zehnder Modulator (MZM) modulates the upstream data. Once the upstream signals are processed and transmitted from the ONU/BS, the uplink is demultiplexed in the remote node in order to extract the P2P wavelength while the uplink wavelength is transmitted to the CS. For the P2P physical connectivity the optical switch maps the corresponding AWG inputs in order to route the P2P wavelength to the right ONU/BS. A frame on the top of Fig. 2(a) highlights the enabling system for the P2P connectivity. The scalability of the proposed router is described in the following discussion. Consider an access network where all the ONU/BSs feature P2P capabilities and the extra capacity channels (N) must be dynamically allocated to (M) ONU/BSs, assuming that each ONU/BS has one fixed optical wavelength and has the possibility of increasing their capacity by a number of extra channels, the number of ports (PAWG) required by the AWG in the RN is given by:
With the number of AWG ports used for P2P purposes given by:
That is, [M-1] input ports more are needed in the AWG to implement the P2P functionality while the [M] output ports are reused due to the cyclical response of the AWG. Following the same considerations, the number of input and output ports required in the optical switch for the P2P connectivity is:
Figure 2(b) shows two scenarios for P2P connectivity among four ONU/BSs. For the scenario shown in Fig. 2(b) left, each ONU/BS has its own fixed channel, the extra capacity channels have been dynamically assigned to ONU/BSs-1 and 3 and ONU/BS-1 has a P2P session with ONU/BS-2 and ONU/BS-3 onto λ1’ = 1561.14 nm, which is the assigned P2P wavelength for sessions originated in ONU/BS-1. Figure 2(b) right, shows a second scenario with a P2P session originated in ONU/BS-3 and transmitted onto its P2P wavelength at λ3’ = 1562.74 nm and ended up in ONU/BS-1 and ONU/BS-2 respectively.
2.2 Wavelength Division Multiplexing (WDM) architecture with wavelength reuse (1:1)
From the techno-economic point of view, the scalability may be an issue in the previous architecture when the access network becomes large due to the tunable laser or stack of lasers placed in the remote node to provide the P2P wavelength. An upgrade of the previous approach corresponds to the architecture shown in Fig. 3(a) . This approach reuses the fixed downlink wavelength assigned to each ONU/BS to enable the P2P sessions.
Similarly to the previous architecture, in this approach the P2P sessions are also served in a one at a time basis (1:1) which results in the same router dimensioning as described in Eq. (1), (2) and (3), but it differs from the previous approach because such 1:1 connectivity can be made in parallel among two or more ONU/BSs by WDM techniques. Unlike the previous configuration, there is no dedicated P2P wavelength but the downlink wavelength is reused to convey both P2P and uplink. The wavelength reuse is performed by the rewriter described in the previous subsection as seen in inset of Fig. 3(a). Thus, upstream data and P2P session share the same wavelength and any tunable laser in the remote node to generate P2P wavelengths is no longer needed. Once the upstream signals are processed and transmitted from the ONU/BS, a coupler splits the uplink wavelength in the remote node in order to extract the P2P content. One of the branches carries the uplink to the CS and the second branch is connected to the optical switch to route the P2P sessions to the corresponding ONU/BSs via the AWG. A frame on the top of Fig. 3(a) highlights the enabling system for the P2P connectivity.
Figure 3(b) shows two scenarios for P2P connectivity among four ONU/BSs featuring wavelength reuse. For the scenario shown in Fig. 3(b) left, each ONU/BS has its own fixed channel, the extra capacity channels have been dynamically assigned to ONU/BSs-1 and 3 and ONU/BS-1 has a P2P session with ONU/BS-2 and ONU/BS-3 onto 1546.64 nm, which is the fixed downlink wavelength for ONU/BS-1. Figure 3(b) right, shows a second scenario with the fixed and dynamic channels assigned in the same way as in the previous scenario and a P2P session from ONU/BS-3 to ONU/BS-1 and ONU/BS-2 using the downlink wavelength for ONU/BS-3 at 1548.24 nm.
2.3 WDM architecture with wavelength reuse (1:M)
The architecture showed in Fig. 4(a) follows the same philosophy as the previous one with regard to the wavelength reuse of the downlink channel and sharing of the reused wavelength for both uplink and P2P sessions. The difference lies on the WDM based broadcasting of P2P sessions which results in a uniform and parallel distribution of P2P content among different ONU/BSs. Fixed wavelengths and extra capacity channels for dynamic reconfigurability are also considered in this approach.
Once the downstream is distributed and the upstream signals are processed and transmitted from a given ONU/BS, a coupler splits the uplink wavelength in the remote node in order to extract the P2P wavelength. Similarly to the previous configuration one of the branches carries the uplink to the CS and the other branch is fed backwards and switched to an optical coupler which outputs map the corresponding AWG input ports. In accordance with the AWG response, input port and wavelength value, the P2P content is broadcasted among different ONU/BSs at the same time. In this approach, this optical switch enables the blocking of non-P2P wavelength sessions. A frame on the top of Fig. 4(a) highlights the enabling system for the P2P connectivity.
Assuming a maximum connectivity capacity (L) defined as the possibility to link one of the (M) ONU/BSs with the rest (M-1) ONU/BSs, the number of ports (PAWG) required by the AWG in the RN is given by:
With the number of AWG ports for P2P determined by:
Figure 4(b) shows two scenarios for P2P connectivity among four ONU/Bs featuring wavelength reuse. For the scenario shown in Fig. 4(b) left, each ONU/BS has its own fixed channel, the extra capacity channels have been dynamically assigned to ONU/BS-1 and 3 and ONU/BS-1 has a P2P session with ONU/BS-2, ONU/BS-3 and ONU/BS-4 onto 1546.64 nm, which is the fixed downlink wavelength for ONU/BS-1. Figure 4(b) right shows a second scenario with a similar allocation for fixed and dynamic channels as in the previous scenario and a P2P session from ONU/BS-3 to ONU/BS-1, ONU/BS-2 and ONU/BS-4 using the downlink wavelength for ONU/BS-3 at 1548.24 nm.
3. Performance evaluation
This section presents the performance evaluation of the described P2P architectures. In all cases the signal degradation on relevant scenarios has been measured in order to assess the performance of the system. The metric used for the evaluation is the signal quality of the P2P connections and uplink. P2P corresponds to all channels that are transmitted among ONU/BSs whereas the uplink corresponds to the signals transmitted from the ONU/BS to the CS.
3.1 TDM architecture with dedicated wavelength (1:1)
For the dedicated wavelength architecture the P2P session between ONU/BS-1 and ONU/BS-2 and 3 has been measured. It corresponds to the scenario shown in Fig. 2(b) left. Figure 5(a) shows the signal quality in ONU/BS-2 and 3 respectively. The results show that the signals for the P2P session with ONU/BS-2 present a penalty of 0.4 dB compared to the back to back curve whereas ONU/BS-3 is penalized in 0.5 dB. Degradation of the QPSK signal shows a low Error Vector Magnitude (EVM) below 3.5% for received optical powers under −24 dBm and with a degradation of roughly 0.1% compared to the back to back value, the results are shown in Fig. 5(b). As for the quality performance for the uplink signals, the results in Fig. 5(c) show a penalty of 0.8 dB for the P2P session originated in ONU/BS-1 (λ19) and 0.9 dB for the P2P session from ONU/BS-3 (λ21) for 1x10−12 BER. In both cases compared to their corresponding back to back curve.
Figure 5(d) shows the degradation of the uplink RF service, with EVM values below 4.6% for received optical powers under −24 dBm featuring a degradation of 1.15% compared to the back to back value. Underlying penalties in both scenarios arise from inherent insertion losses of the optical components while larger penalties in the uplink RF signals come from the crosstalk of the remaining downlink wavelength due to the reuse of the channel.
3.2 WDM architecture with wavelength reuse (1:1)
The wavelength reused architecture has been evaluated with the scenario shown in Fig. 3(b), left which corresponds to the P2P session from ONU/BS-1 and ONU/BS-2 and 3 using the fixed wavelength assigned to ONU/BS-1 (λ1 = 1546.65). Figure 6(a) shows the signal quality in ONU/BS-2 and 3 respectively and additionally it also shows the BER curve for the downstream signal at ONU/BS-1.
The results show that the signals for the P2P session with ONU/BS-2 and ONU/BS-3 present a penalty of 1.1 dB compared to the back to back curve. Degradation of the QPSK signal shows an EVM below 4.8% for received optical powers under −24 dBm for ONU/BS-2 and 3 and with a degradation of roughly 1.3% compared to the back to back value as seen in Fig. 6(b). As for the quality performance for the uplink signals, the results in Fig. 6(c) show a penalty of 0.85 dB for the P2P session from ONU/BS-1 to ONU/BS-2 and 3 for 1x10−12 BER. Figure 6(d) shows the degradation of the uplink RF service, the results show EVM values below 4.65% for received optical powers under −24 dBm featuring a degradation of 1.2% compared to the back to back value. Negligible penalties undergo the baseband signal in ONU/BS-2 and 3 with respect to the signals received in ONU/BS-1 (~0.6 dB) and roughly 1.1% for the RF services.
3.3 WDM architecture with wavelength reuse (1:M)
The system performance was evaluated by measuring the quality of signals in the scenario shown in Fig. 4(b) left, which corresponds to the P2P session from ONU/BS-1 to ONU/BS-2, 3 and 4. Figure 7(a) shows the signal quality in ONU/BS-2, ONU/BS-3 and ONU/BS-4 respectively and additionally it also shows the BER curve for the downstream signal at ONU/BS-1. The results show that the signals for the P2P session with ONU/BS-2, ONU/BS-3 and ONU/BS-4 present a penalty of 1.1 dB compared to the signal in ONU/BS-1 and 1.5 dB penalty with respect to the back-to-back curve.
Degradation of the QPSK signal shows an EVM below 4.8% for received optical powers under −24 dBm for ONU/BS-2, ONU/BS-3 and ONU/BS-4 and with a degradation of roughly 1.1% compared to the quality in ONU/BS-1 and 1.3% with respect to the back to back value as seen in Fig. 7(b). As for the quality performance for the uplink signals, the results in Fig. 7(c) show a penalty of 1.1 dB for 1x10−12 BER for the signals transmitted to the CS from ONU/BS-1 and ONU/BS-3. Figure 7(d) shows the degradation of the uplink RF service, the results show EVM values below 4.75% for received optical powers under −24 dBm featuring a degradation of 1.25% compared to the back-to-back value.
Three different architectures of a wavelength router based remote node to enable P2P connectivity in a converged wired/wireless RoF access network were presented. The first architecture uses a tunable laser in the remote node to generate the P2P wavelength while the second and third approach reuses the fixed wavelength of the ONU/BS to enable the P2P session. Remarkable differences among the systems lies in the capability of holding simultaneous P2P sessions, while the first two approaches allows 1:1 connectivity based on TDM and WDM respectively, the third one enable WDM based 1:M assignment. The approach enables the broadcasting of P2P sessions among different ONU/BSs whereas the demonstrated functionalities negotiate the backhaul requirements for 4G radio access in terms of capacity, multi-service capability and inter ONU/BS logical mesh connectivity.
A discussion on the RN dimensioning was also presented showing that, in general, the requirements of the AWG capacity increase by a factor of (M-1) ONU/BSs while the size of the optical switch is directly proportional to the number of ONU/BSs (M) featuring P2P connectivity.
The experimental measurements confirm the good performance of the 1:1 systems with no noticeable degradation effect on the transported signals, 0.2% degradation for EVM and 0.7 dB penalties in average for 1x10−12 BER in the first architecture whereas 1.2% degradation for EVM with 1.1 dB penalty in average for 1x10−12 BER in the second architecture. Despite a slight better performance of the dedicated wavelength architecture with respect to the wavelength reused one, the use of the tunable laser may arise techno-economic issues that may result in a limitation of the access networks scalability. As for the 1:M approach the experimental measurements confirm the good performance of the system with no noticeable degradation effect on the transported signals, 1% degradation for the EVM and 1 dB penalty in average for 1x10−12 BER in both downlink and uplink. Negligible penalties undergo the baseband signal in ONU/BSs 2, 3 and 4 with respect to the signals received in ONU/BS-1 (~0.6 dB) and roughly 0.8% for the RF services.
The research leading to these results has received funding from the European Community’s Seventh Framework Program (FP7) under project 212 352 ALPHA “Architectures for fLexible Photonic Home and Access networks” and Generalitat Valenciana through the PROMETEO research excellency award programme GVA PROMETEO 2008/092.
References and links
1. A. Sentinelli, G. Marfia, M. Gerla, L. Kleinrock, and S. Tewari, “Will IPTV ride the peer-to-peer stream?” IEEE Commun. Mag. 45(6), 86–92 (2007). [CrossRef]
2. C. P. Larsen, A. Gavler, and S. Junique, “The Impact of Peer-to-Peer networking on user behaviour and network design,” in Optical Fiber Communication Conference and Exposition (OFC), Optical Society of America, Washington, DC, 2010, paper NTuA5.
3. G. K. Venkatesan, and K. Kulkarni, “Wireless backhaul for LTE - requirements, challenges and options,” 2nd International Symposium on Advanced Networks and Telecommunication Systems, 2008. ANTS '08, pp.1–3, 15–17 Dec. 2008.
4. J. J. Vegas Olmos, T. Kuri, T. Sono, K. Tamura, H. Toda, and K. I. Kitayama, “Wireless and Optical-Integrated Access Network With Peer-To-Peer Connection Capability,” IEEE Photon. Technol. Lett. 20(13), 1127–1129 (2008). [CrossRef]
6. G. Puerto, J. Mora, B. Ortega, and J. Capmany, “Wavelength Data Rewriter for Centralized-Source Radio-Over-Fiber Access Networks,” IEEE Photon. Technol. Lett. 22(15), 1102–1104 (2010). [CrossRef]