Performance Comparison of Reservation MAC Protocols for Broadband Powerline Communications Networks Halid Hrasnica*, Abdelfatteh Haidine, Ralf Lehnert Chair for Telecommunications, Dresden University of Technology D-01062 Dresden, Germany
ABSTRACT We study the MAC layer of powerline communications (PLC) transmission systems applied to telecommunication access networks. PLC networks have to operate with a limited signal power which makes them more sensitive to disturbances from the electrical power supply grid and from the network environment. Well-known error handling mechanisms can be applied to the PLC systems to solve the problem of transmission errors caused by the disturbances (e.g. FEC and ARQ). However, the use of this mechanisms consumes a part of the transmission capacity and therefore decreases the already limited net data rate of the PLC systems. Because of the limited bandwidth, PLC networks have to provide a very good network utilization. Also a sufficient QoS is required, which can be reached by usage of efficient methods for the network capacity sharing – MAC protocols. The impulsive noise influences very much the error-free transmission. Therefore, this investigation includes modeling of several disturbance scenarios, too. We propose the reservation MAC protocols to be applied to the PLC access networks, because they are suitable to carry hybrid traffic with variable data rates ensuring a high network utilization. The analysis of the basic reservation protocols shows that the ALOHA random protocol can not deal with frequent transmission demands but it is more robust against disturbances than the polling based access protocol. The ALOHA protocol can be improved by the piggybacking method which degrades the collision probability and accordingly, shorts access delay. The polling protocol is extended with insertion of a contention component building a hybrid access method, which makes the access delays shorter, if there is a small number of stations in the network. Generally, the problems caused by the frequent transmission requests remain in all investigated access methods. However, the ALOHA based protocols show worst behavior in this case. Keywords: Powerline communications, MAC, reservation, modeling, simulation, performance analysis, disturbances
1. INTRODUCTION This work is a continuation of the investigations presented in 1 and 2 considering the MAC layer in the Powerline Communications (PLC) access networks. In 1 we presented the simulation model for the investigations of the PLC MAC layer including the simulation results for the first two variants of the reservation protocol for PLC; an ALOHA based random access protocol and a polling based dedicated access protocol. Various solutions for the protocol optimization and their improvement are discussed in 2 and in this investigation, we analyze and compare an ALOHA based reservation protocol with piggybacking, as an extension of the basic ALOHA protocol and a hybrid polling protocol with a random component. We consider PLC access systems using the low-voltage power supply network for the connection of end-users/subscribers to a wide area network as an alternative solution for the communication last mile 3, 4. There is a number of subscribers in a low-voltage electrical power supply network who have to share the transmission capacity of a PLC access network. Therefore a high gross data rate on the medium is necessary to ensure a sufficient QoS and to make PLC systems competitive to the other access technologies (xDSL, usage of CATV, WLL, ...). PLC systems applied to the telecommunication access networks use a frequency spectrum of up to 30 MHz and act as antenna producing electro-magnetic radiation, which causes disturbances to other telecommunication services working in *
Corresponding author – Email:
[email protected], Internet: http://www.ifn.et.tu-dresden.de/~hrasnica, Telephone: +49 351 463-3474, Fax: +49 351 463-7163
this frequency range. Because of that, PLC networks have to work with a limited signal power which makes PLC systems more sensitive to disturbances from the electrical power supply network and from the PLC network environment. Wellknown error handling mechanisms can be applied to the PLC systems to solve the problem of transmission errors caused by the disturbances (e.g. FEC – Forward Error Correction and ARQ – Automatic Repeat reQuest mechanisms). However, the use of this mechanisms consumes a part of the transmission capacity (overhead, retransmission) and therefore decreases the already limited data rate of the PLC systems. Because of the shared transmission medium, PLC networks have to provide a very good network utilization keeping also a sufficient QoS, which can be reached by usage of efficient methods for the network capacity sharing – Media Access Control (MAC) protocols. The paper is organized as follows: First, we describe briefly the investigated PLC access system as well as applied simulation and disturbance model (section 2). After that, we analyze possible solutions for the PLC MAC layer and discuss several variants of the reservation MAC protocols (section 3). Further, a comprehensive simulation based protocol performance analysis is given. We compare ALOHA access protocol and its extension with piggybacking method (section 4). In section 5, a hybrid protocol with mixed polling and random access protocol is compared with a polling method.
2. PLC SYSTEM FEATURES AND MODELING 2.1.
Network Structure
PLC access networks are connected to the backbone communication networks via a transformer station, or via any other station in the network, building a physical tree structure 3, 4, 5. Independently of the PLC network topology the communication between the users of a PLC network and a wide area network (WAN) is carried out over a base station, normally placed in the transformer unit (Figure 1).
Figure 1 : Logical Bus Topology of the PLC Network A transmission signal sent by the base station in the downlink direction (from the base station to the users) is transmitted to all network subsections. Therefore, it is received by all users in the network. In the uplink direction (from the users to the base station), a signal sent by an user is transmitted not only to the base station, but also to all other users in the network. That means, the PLC access network holds a logical bus structure in spite of the fact that the low-voltage supply networks have physically a tree topology 4. 2.2.
PLC Services
We consider the PLC transmission system as a bearer service carrying teleservices, which make possible usage of various communication applications 6. The MAC protocol to be implemented in a PLC system has to provide features for realization of different teleservices 1, 2, 3, 4 ensuring the needed QoS. On the first place, the classical telephone service and internet based data transmission have to be provided by the PLC access systems, because of their importance and their big penetration in the communication world. 2.3.
Transmission System
We consider an OFDM (Orthogonal Frequency Division Multiplexing) based transmission system using a number of subcarriers distributed in a frequency spectrum 1, 4. Each sub-carrier has a transmission capacity and it is possible to make a groups of the sub-carriers to build up transmission channels with a higher capacity 7. We assume that a transmission channel offers a fixed data rate of 64 kbps like in the investigation presented in 1.
However, in any transmission technique we find a structure consisting of a number of transmission channels divided in the frequency (OFDMA or FDMA) , time (TDMA) or code (CDMA) domain. So, we can conclude that the PLC transmission systems seem to have a logical channel structure independently of the used transmission technology 2, 4. The logical transmission channels have different meaning for each considered transmission method, but the investigations, done on the logical level, can be applied to any of the transmission schemes (e.g. OFDMA investigation to TDMA, ...).
2.4.
Simulation Model
The simulation model, also used in the previous investigation 1 and 4, consists of a number of bidirectional transmission channels (Figure 2) with a fixed data rates which connect the network users with a base station. There is a possibility for modeling various disturbance types which make the transmission channels unavailable according to the chosen disturbance scenarios. The disturbances can be modeled to affect both single and multiple transmission channels representing an impact of disturbances in the frequency spectrum.
Base Station
Figure 2 : Generic Simulation Model Each user can be modeled to provide multiple telecommunication services (e.g. telephony and Internet). The user model to includes the implementation of a considered MAC protocol. Additionally, model of the base station should include a traffic scheduling strategy which is applied in the network providing all features of a request-answer procedure. Both user and base station implementations have to realize an error handling mechanism (e.g. ARQ), case it is applied. 2.5.
Disturbance Modeling
As shown in 1 a transmission channel can be modeled by a Markov chain with two states: • •
OFF - the channel is disturbed, no transmission is possible ON - the channel is available
These two states can be modeled by two random variables which represent interarrival times of the impulses/disturbances, moving a channel into the state OFF, and the disturbance duration. Both random variables are negative exponentially distributed 8, 9. In this investigation we simulate PLC networks with the following three disturbance scenarios: • •
Disturbance-free network Lightly disturbed network - 200 ms interarrival time of the impulses/disturbances
•
Heavily disturbed network - 40 ms interarrival time of the disturbances
We set the mean duration of a disturbance impulse to 100 µs and assume that the noise impulses with duration shorter than 300 µs do not cause errors 1, 2, 4. The disturbances are modeled independently for each transmission channel.
3. ANALYSIS OF THE PLC MAC LAYER 3.1.
MAC Protocols for PLC
A MAC protocol specifies a resource sharing strategy – access of many users to the network transmission capacity – applied to a multiple access scheme 1, 4, 10. Fixed access strategies are suitable for continuous traffic, but not for bursty traffic which is typical for data transfer provided in the PLC access networks. Dynamic access schemes are adequate for data transmission and in some cases it is possible to ensure a satisfactory transmission quality for delay-critical traffic. Dynamic protocols with contention can not ensure any guarantees of QoS for time-critical services and also 100% network utilization can not be reached. Collision-free dynamic protocols can be realized using token passing, polling or reservation methods. Token passing and polling make possible realization of some QoS guarantees in the network. However, with an increasing number of network stations the time between two sending rights for a stations (round-trip time of tokens or polling messages) becomes longer, making both protocols not suitable for time-critical services 1, 4. In the case of reservation protocols, a kind of pre-reservation of the transmission capacity for a particular user is done. A transmission request is submitted by user to a central network unit (e.g. base station in PLC network) using either a fixed or a dynamic access schemes. Transmission systems with the reservation access scheme are suitable to carry hybrid traffic (mix of traffic types caused by various services) with variable transmission rate 10. Satisfaction of various QoS requirements is also possible and a good network utilization can be reached. Because of that we propose the application of the reservation protocols in the PLC access networks. 3.2.
Analysis of Basic Reservation Protocols
In 1 we investigate two different protocol solutions for the signaling channel: •
ALOHA - a contention protocol
•
Polling protocol - with dedicated reservation
In the first case, according to the ALOHA protocol, a network station tries to send transmission requests/demands, using socalled random request slots (Figure 3), to the base station over a signaling channel and after that waits for an answer from the base station, which includes information about medium access rights for the requested transmission. In the case of collision with a request from an other network station, the both affected stations will try to retransmit their transmission demands after a random time.
Figure 3 : Basic Reservation Protocols In the second case, there is a polling procedure realized by the PLC base station which sends so-called polling messages to each network station according to the Round Robin procedure. A station receiving a polling message has the right to send a
transmission request and afterwards the base station answers with information about medium access rights. The collisions are not possible (Figure 3), but a request can be disturbed and it has to be retransmitted in the next request slot, which is reserved for the station. The user packets are reassembled into the smaller units - PLC segments - with 28 bytes payload length 2, 4. In both ALOHA and polling based protocols the users request a number of PLC segments to be transmitted according to the size of an arrived user packet (e.g. IP packet). A transmission demand arrives the base station if it was not disturbed by disturbance or collision (only for ALOHA principal). If the request was successful, the base station answers with the number of data slots which have to be passed by the station before it starts to send (Figure 4). The passed slots (if any) are used by other stations. The answer from the base station can be also disturbed and in this case the whole requesting procedure is repeated according to the used access method.
Figure 4 : Request - Acknowledgement Procedure The results of the investigation done in 1 can be summarized as follows: •
The access times with ALOHA based protocols are significantly shorter than with polling access method in the case that the transmission requests occur relative seldom with accordingly few number of the collisions on the signaling channel. However, if the collision probability increases (e.g. with increasing network load or number of subscribers in the PLC network), the advantage of the ALOHA based protocol disappears.
•
On the other hand, the ALOHA protocol shows a better robustness against disturbances because of its random mechanism and changes its behavior more seldom than the polling based protocol.
To find a suitable solution for the MAC protocol of the PLC network we investigate possibilities for optimization of the both protocol solutions. The ALOHA based protocol can be improved with a reduction of the collision probability in the signaling channel. The disadvantages of the polling based protocol can be improved by inserting a contention component into the protocol making it more robust against disturbances and decreasing the access times. 3.3.
Investigation of Advanced Reservation Protocols
In the investigation 2 we analyzed the possibilities for the improvement of both ALOHA and polling based reservation protocols. To decrease the collision probability on the signaling channel with the ALOHA access principal, we propose usage of the piggybacking method. A station transmitting the last segment of a packet (e.g. IP packet) can use this segment also to request a transmission for a new packet, if there is one in its queue. In this case the signaling channel is not used for the request and the collision probability decreases. To improve the polling based protocol we add a random component making a hybrid access method 2. In this case, the signaling channel is divided into a random and dedicated part (Figure 5). There are twice fewer number of dedicated request slots than in the polling access method (see above). Firstly, the stations try to make the request in the random slot. If it is not successful, the dedicated slots are used (the slots in the signaling channel addressed particularly to a station).
Figure 5 : Hybrid Access Protocol To investigate protocol improvement within piggybacking ALOHA and hybrid protocols we use the simulation and disturbance models described in section 2. We study a network with a variable number of users (50-500) using data service which represents the internet traffic. Internet users transmit usually in the uplink direction only small packets representing the requests for the download of WWW pages. Because of that we assumed a lower offered traffic load per user which amounts to 25 kbps. We consider two average sizes of user packets which are transmitted in the uplink direction; 1500 and 300 bytes to be able to compare network performance in case of larger and smaller user packets. The packet sizes are negative exponential distributed as well as the interarrival time of the packets. According to the chosen offered traffic load the mean intararrival times come to 4,8 and 0,96 seconds for 1500 and 300 bytes user packets respectively and the offered traffic rate per user is the same for both cases and they can be compared.
4. ALOHA PROTOCOL WITH PIGGYBACKING 4.1.
Network Utilization
We observed network utilization as relation between used network capacity (only error-free segments are taken into account) for the data transmission and the common capacity of the PLC network. If we analyze the networks with an average packet size of 1500 bytes, there is no difference between ALOHA access method and ALOHA with piggybacking access (Figure 6). Maximum network utilization is reached within the network without disturbances (about 93%, remaining 7% reserved for signaling, like in 1). In the lightly disturbed network the maximum utilization amounts to 83% and in the heavily disturbed network it is about 50%. Average packet length: 1500 bytes 1.4 ALOHA not_disturb PB not_disturb ALOHA light_disturb PB light_disturb ALOHA heavy_disturb PB heavy_disturb
1.2
Utilization
1 0.8 0.6 0.4 0.2 0 50
100
150
200 250 300 350 Number of stations
400
450
500
Figure 6 : Average Network Utilization (ALOHA and ALOHA with Piggybacking) In the investigation 2 we conclude that the utilization in the network with smaller average packet length can be improved with the usage of the piggybacking in the ALOHA access method, which decreases the collision probability in the signaling channel (in the case of the smaller packets the transmission requests are more frequent). In the network without
piggybacking maximum utilization is 27% (Figure 7). The usage of the piggybacking improves the utilization up to 31%, also under heavily disturbed conditions (in the network without piggybacking it was 23%). The better utilization of the piggybacking method remains in the whole investigated network load domain (number of stations in the network), in all networks; both networks with disturbances and network without disturbances. However, the reached utilization is not sufficient. Average packet length: 300 bytes 0.35 ALOHA not_disturb PB not_disturb ALOHA light_disturb PB light_disturb ALOHA heavy_disturb PB heavy_disturb
0.3
Utilization
0.25 0.2 0.15 0.1 0.05 0 50
100
150
200 250 300 350 Number of stations
400
450
500
Figure 7 : Average Network Utilization (ALOHA and ALOHA with Piggybacking) 4.2.
Access Delay
We measure the access delay as time needed for realization of the requesting procedure for the transmission of an arrived packet. This procedure includes transmission of a request message to the base station and reception of its answer about the access rights. For both access methods, ALOHA with and without piggybacking, the access delays remain the same in the low network load area (Figure 8 - up). Networks under heavier disturbance conditions have a longer access delays in this load area. Shortly below the point where the networks reaches the maximum utilization (150, 250, 300 stations in heavily, lightly and not disturbed networks respectively; see Figure 6) the access delay decreases significantly in the networks with piggybacking access method. Above this point (near to saturation point), networks have the maximum utilization and the transmission time of the packets (in this case long packets, average size 1500 bytes) increases. Accordingly, there is a higher probability that a new packet arrives to the sending queue, during the transmission of an another packet. That means, the probability that the requests are sent using the piggybacking increases too, which decreases access delays in the higher network load area. In the area of higher network load (above the saturation point) we recognize an effect, that the access delays in the networks under heavier disturbance behavior are shorter, for both access methods with and without piggybacking (Figure 8 - up). If the maximum network utilization is reached, the transmission times of the packets increases as well as the data throughput decreases (see next subsection). Accordingly, the number of the requests decreases (a new request is sent after a packet is successfully transmitted). Also the number of requests using the piggybacking increases (see above), and the collision probability in the signaling channel becomes lower. Because of that, the access delays are shorter in this case. Note, the access delay include only the time needed for the requesting procedure (the transmission time of the packets is not included). In the investigation 2 we expected to decrease very critical access delays in the networks with the shorter packet length by using of the piggybacking. We see (Figure 8 - down) that the access delays decrease with the piggybacking, approximately twice (note, the y axe is logarithmic). However, the access delays are still too long above 150 stations in the network.
Average packet length: 1500 bytes
Access delay / ms
1000 ALOHA not_disturb PB not_disturb ALOHA light_disturb PB light_disturb ALOHA heavy_disturb PB heavy_disturb
100
10 50
100
150
200 250 300 350 Number of stations
400
450
500
Average packet length: 300 bytes 1e+08 ALOHA not_disturb PB not_disturb ALOHA light_disturb PB light_disturb ALOHA heavy_disturb PB heavy_disturb
1e+07
Access delay / ms
1e+06 100000 10000 1000 100 10 50
100
150
200 250 300 350 Number of stations
400
450
500
Figure 8 : Mean Access Delay (ALOHA and ALOHA with Piggybacking) 4.3.
Data Throughput
Like in the investigation 1, the data throughput behavior follows the network utilization. If we consider the networks with the packet size of 1500 bytes, there is no difference between the networks with and without the piggybacking (Figure 9). On the other side, the improvement of the network utilization reached with the piggybacking in the case of the smaller packets (300 bytes) is visible.
Average packet length: 1500 bytes
Average packet length: 300 bytes 1
ALOHA not_disturb PB not_disturb ALOHA light_disturb PB light_disturb ALOHA heavy_disturb PB heavy_disturb
1.2
0.8
Throughput
Throughput
1
ALOHA not_disturb PB not_disturb ALOHA light_disturb PB light_disturb ALOHA heavy_disturb PB heavy_disturb
0.8
0.6
0.4
0.6 0.2 0.4 0 50
100
150
200 250 300 350 Number of stations
400
450
500
50
100
150
200 250 300 350 Number of stations
400
450
500
Figure 9 : Average Relative Data Throughput (ALOHA and ALOHA with Piggybacking)
5. HYBRID POLLING-ALOHA ACCESS PROTOCOL 5.1.
Network Utilization
In the case of the network with the larger size of the packets, the network utilization remains the same for both used access protocols; polling and hybrid protocol. The results correspond also to the both variants of ALOHA protocols (with and without piggybacking, see Figure 6) in the case of the larger packets as well as to the investigations in 1. Average packet length: 300 bytes 1.2 Poll not_disturb Hyb not_disturb Poll light_disturb Hyb light_disturb Poll heavy_disturb Hyb heavy_disturb
Utilization
1
0.8
0.6
0.4
0.2 50
100
150
200 250 300 350 Number of stations
400
450
500
Figure 10 : Average Network Utilization (Polling and Hybrid access protocols) If we observe the networks with the smaller average size of the packets (Figure 10), we find a lower network utilization in the case of the networks with the hybrid access protocol. This difference increases with the higher number of stations in the network. In the higher network load area, the collisions, which are possible in the contention part of the signaling channel provided by the hybrid protocol, occur more frequently. Note, that within hybrid protocol there are twice slots for the dedicated access fewer. In the case of a collision, the stations wait for the dedicated request slot and network utilization decreases. However, the utilization remains much better than in both investigated ALOHA protocols (section 4) for smaller packets.
5.2.
Access Delay
The insertion of a random component into the polling based access protocol should decrease the access delays, especially in the networks with the smaller number of stations 2. As we see (Figure 11 - up), there is a significant decrees of the access delay in this network load area in the case of the networks with larger packets. The access delays come nearer to the values measured with polling method for larger number of the stations in the network. In spite of the improvement for the access delays, they remain longer than in the networks with the both ALOHA access methods (Figure 8 - up). The effect that the heavier disturbed networks have a bit shorter access times, recognized in the networks with ALOHA access protocols, can be also seen in the networks with the hybrid protocol (Figure 11 - up) while the polling protocol does not show this feature.
Average packet length: 1500 bytes 100000 Poll not_disturb Hyb not_disturb Poll light_disturb Hyb light_disturb Poll heavy_disturb Hyb heavy_disturb
1000
100
50
100
150
200 250 300 350 Number of stations
400
450
Average packet length: 300 bytes 100000 Poll not_disturb Hyb not_disturb Poll light_disturb Hyb light_disturb Poll heavy_disturb Hyb heavy_disturb
10000
Access delay / ms
Access delay / ms
10000
1000
100
50
100
150
200 250 300 350 Number of stations
400
450
500
Figure 11 : Mean Access Delay (Polling and Hybrid access protocols)
500
In the case of the networks with shorter packets (Figure 11 - down), the contention component of the hybrid protocols increases the access delays. Only for very small number of stations (50), we recognize an improvement of the access delays. A common number of the requesting slots in the hybrid protocol is divided in two groups: random and dedicated slots 2. Because of that, there are twice dedicated slots fewer than in the polling protocol and other half of the slots can be accessed randomly. The investigation in 1 and in the section 4 show unstable behavior of the random protocol if the small packets are transmitted and number of collisions as well as the access delays increase significant and the random slots are barely used for the successful request. Accordingly, the fewer number of the dedicated slots has to take over collided requests and that causes the delay increase. 5.3.
Data Throughput
The results for the data throughput follow the behavior of the network utilization (like in the investigation 1 and section 4). So, in the case of the larger packets (1500 bytes) there is no difference between polling and hybrid access protocol (see below). In the case of the smaller packets (300 bytes), hybrid protocol shows lower data throughput according to the reached lower network utilization (Figure 10).
6. CONCLUSIONS We investigate improvements of two basic solutions for the reservation MAC protocols to be applied in the broadband PLC networks. A random ALOHA reservation protocol is extended with the usage of piggybacking and a contention component is added to a dedicated polling access protocol making a hybrid access method. The protocols are simulated under different disturbance conditions; heavily and lightly disturbed and not disturbed networks. The simulation were done for longer user packets - less intensity of the transmission requests - and for shorter packets - frequent requests. The protocol extensions did not bring any changes in the networks with longer user packets, if we consider network utilization and data throughput. For shorter packets, protocol with piggybacking shows better utilization and throughput, but still much lower than in the networks with polling and hybrid access methods. The hybrid protocol decreases network utilization and data throughput compared with the polling. Piggybacking decreases access delay, especially if the network is highly loaded. That leads to recommendation of its implementation not only in the ALOHA protocol but also in polling or hybrid access method, too. The hybrid protocol decreases access delays of the polling method, particularly for smaller number of stations in the network. However, if we consider the networks with smaller user packets, this improvement exists only for a very small number of the stations. The networks with smaller user packets and accordingly, frequently transmission requests still have the long access delays. The improvement is visible with the insertion of the piggybacking and partly with usage of the hybrid random/dedicated protocols, but the access times go beyond one second, which is too long (300 bytes long packets need for the transmission 4 ms in the investigated networks). On the other side, in the networks with the short packets it is not possible to reach a full or nearly full network utilization and data throughput. In our future work we will investigate further solutions to overcome problem of very frequently transmission requests in a per-packet reservation MAC protocol for PLC.
ACKNOWLEDGMENTS The authors thank to ONELINE, Barleben, Germany, for supporting our investigations. Part of this work has been done within the EU project PALAS (Powerline as an Alternative Local AcceSs).
REFERENCES 1.
2.
H. Hrasnica, A. Haidine, Modeling MAC Layer for Powerline Communications Networks, Internet, Performance and Control of Network Systems, part of SPIE's symposium on Information Technologies, 5-8 November, 2000; Boston MA, USA H. Hrasnica, A. Haidine, R. Lehnert, Reservation MAC Protocols for Powerline Communications, ISPLC2001 - 5th International Symposium on Power-Line Communications and its Applications, 4 - 6 April 2001 - Malmö, Sweden
3.
H. Hrasnica, R. Lehnert, Powerline Communications in Telecommunication Access Area (Powerline Communications im TK-Zugangsbereich), VDE World Microtechnologies Congress - MICRO.tec 2000 - ETG-Fachtagung und -Forum: Verteilungsnetze im liberalisierten Markt - September 25-27, 2000 - Expo 2000, Hannover, Germany 4. H. Hrasnica, R. Lehnert, Powerline Communications for Access Networks – Performance Study of the MAC Layer, III International Conference on Telecommunications BIHTEL 2000 “Telecommunication Networks” - October 23-25, 2000 - Sarajevo, Bosnia and Herzegovina 5. O. G. Hooijen, On the Channel Capacity of the Residential Power Circuit used as a Digital Communications Medium; IEEE Communications Letters, Vol. 2, No. 10, October 1998 6. ITU-T Recommendation I.210 (03/93): ISDN Service Capabilities 7. B. Walke, Mobilfunknetze und ihre Protokolle – Band 1 (in German), B. G. Teubner - Stuttgart, Germany, 1998 (ISBN 3-519-06430-8) 8. M. Zimmerman. K. Dostert, The Low Voltage Power Distribution Network as Last Mile Access Network – Signal Propagation and Noise Scenario in the HF-Rang, International Journal of Electronics and Communications (AEU) 54 (2000) N°. 1, 13-22 9. M. Zimmerman, K. Dostert, An Analysis of the Broadband Noise Scenario in Powerline Networks, 2000 International Symposium on Powerline Communications and its Applications – April 2000 – Limerick –Ireland 10. I. F. Akyildiz, J. McNair, L. C. Martorell, R. Puigjaner, Y. Yesha: Medium Access Control Protocols for Multimedia Traffic in Wireless Networks. IEEE Network - July/August 1999
