The use of spacecraft for data relay to provide radio coverage to the entire Earth dates back to 1945, when Arthur C. Clarke, at that time serving in the Royal Air Force as officer and radar specialist, published the visionary article in Wireless World entitled “Extra-Terrestrial Relays. Can Rocket Stations Give World-wide Radio Coverage?” Twelve years later, the first communications satellite, named Sputnik 1, was launched by the Russians. It carried two radio transmitters at 20.005 and 40.002 MHz, with batteries lasting 22 days. Today, the number of communications satellites listed in the National Space Science Data Center (NSSDC) Master Catalog is greater than 2000; the number of active communication satellites in geosynchronous orbits, characterized by an orbital period equal to one sidereal day, is almost 400; and the lifetimes of such satellites are typically around 15 years. Television and radio broadcasting are by far the most important markets for satellite manufacturers and operators, accounting today for almost 80% of the total satellite service revenue. In contrast, for many years, bidirectional communication services via satellite remained a relatively small, although important, market niche, mainly addressing institutional or professional applications. This was chiefly due to the high cost of satellite airtime and terminals and to the high latency due to the propagation delay. Only recently, affordable broadband-access satellite networks have been emerging in the United States, complementing the terrestrial networks' coverage in locations where wireline solutions are not economically viable. At the time of this writing (2014), Google has announced plans to invest in a fleet of satellites that will expand Internet access to unconnected regions of the world.
Use of satellite communication for M2M services
Moreover, an increasing number of emerging applications are based on the sporadic transmission of short messages from and to remote sensors or mobile devices used to track specific events or monitor automatic systems, and they might radically change the current picture. These applications are typically labelled under the broad term “machine-to-machine” (M2M), and include, for example, industrial supervisory control and data acquisition (SCADA) systems; fleet management or containers' tracking; public systems, such as automatic highway tollgates and traffic-light controllers; and energy systems, such as current sensors in a solar panel array and water-level sensors in a dam. Although terrestrial wireless networks are expected to be capable of serving most of the traffic generated by such applications, the intrinsic cross-border nature of satellite communications makes it the ideal complement to provide M2M services over a wide area to remote and sparsely populated locations.
This article describes a solution for M2M applications called S-band Mobile Interactive Multimedia (S-MIM), recently standardized by the European Telecommunications Standards Institute (ETSI). S-MIM (Fig. 1) is an integrated satellite-terrestrial mobile system capable of providing interactive broadcasts and multicasts as well as packet data acquisition services to subscribers. The technical solution adopted for its satellite uplink channel (that is, from terminal to gateway) is particularly innovative and offers a low-cost yet spectrally efficient solution to send short messages with modest power requirements on the terminal side via a geosynchronous satellite acting as relay. Remarkably, the consequent reduction of capital and operational expenditures is expected to significantly contribute to the enlargement of market share for satellite-based M2M services. The rest of this article focuses on the uplink transmission scheme.
Whenever a certain number of terminals share a physical wireless communication channel (that is, a portion of the available radio spectrum), a suitable multiple-access protocol is required. Existing multiple-access protocols can be grouped into two main categories, namely reservation-based and contention-based. In the first case, often referred to as demand assignment multiple access (DAMA), coordination among all terminals is required to ensure that each one has access to an exclusive portion of the available spectral and time resources. In the simplest cases, this can be achieved, for example, by partitioning the available radio spectrum into smaller frequency subchannels or different time slots. A central network control unit will dynamically allocate available time and frequency resources to the requesting terminals. In contrast to this procedure, contention-based access protocols assume that no, or very little, coordination exists among terminals. This decentralized approach does not guarantee that data packets transmitted by different terminals will not overlap in time or frequency, thus leading to mutual interference (Fig. 2). The main advantage is that the network does not require the extra burden (in time and signaling overhead) of centralized resource management.
The simplest type of contention-based protocol is named ALOHA. This was a pioneering computer networking system developed at the University of Hawaii in the 1960s to interconnect by radio links computer terminals located in the Hawaiian Islands to the university campus mainframe. The basic version of the ALOHA protocol consists in letting each terminal transmit at arbitrary time instants, regardless of what other terminals sharing the same channel will do. Communication theory tells us that, when the power of the received packets is balanced (that is, when each terminal sets its transmitted power to a level that ensures that all packets are received with the same power level), the maximum fraction of the channel traffic load that could be successfully received, the so-called throughput, is roughly 0.184 with ALOHA. [In the following, we will measure the throughput as the ratio of information bits per signalling unit (chip), that is, in bits/chip. This is a more general and accurate way to measure the spectral efficiency of a random-access scheme.] This value of the throughput can be increased to 0.368 by introducing time slots and allowing terminals to send their packets only at the beginning of a slot; the resulting protocol is called slotted ALOHA. (In practice, the achievable random-access throughput is much lower than its peak value; in fact, it may be two orders of magnitude less. This is because, to minimize retransmission over the large-latency satellite channel, the packet loss ratio, defined as the ratio of total lost packets to total transmitted packets, should be kept below 10−3.) It should be noted that this modification already implies a higher level of coordination among all the terminals, in order to keep the transmitted packets synchronized to the common network timing reference for aligning each packet to be transmitted at the beginning of a time slot. In general, any type of coordination among terminals requires exchanging signaling information over the communication network and, in many cases, the presence of a central entity in charge of gathering, processing, and distributing this information is also required. In a properly designed communication network, the amount of exchanged data signaling should be minimized to avoid subtracting precious spectral resources from the transmission of real data traffic. The consequence of poor throughput performances of ALOHA-like protocols is that, today, they are mainly used only during satellite terminal logon to exchange initial signaling, whereas reservation-based access protocols are preferred for traffic data exchange.
Use of SSA with iSIC for M2M applications
However, in the case of networks with a very large number of terminals characterized by very sporadic traffic activity, as for the emerging M2M applications described earlier, the signaling overhead associated with the most common reservation-based access protocols becomes an intolerable source of inefficiency and negatively affects the network scalability. Based on the previously stated consideration, the access protocol used in the uplink of the S-MIM system is based on an advanced type of random-access scheme. This approach overcomes the aforementioned ALOHA throughput limitation, while preserving its main advantage with respect to reservation-based access protocols, namely, a very simple logon procedure and very low signaling overhead for large networks. Spread-spectrum ALOHA (SSA), together with the application at the demodulator of a technique called iterative successive interference cancellation (iSIC), provides the best fit to this application scenario. The basic concepts of SSA and iSIC will now be described.
Conventional SSA allows the resolution of packet collisions thanks to the terminal-unique spreading sequence discrimination characteristics. (In reality, each terminal is reusing the same spreading sequence but with a different time, frequency, and power offset so as to make it appear unique at the gateway demodulator.) Relatively high throughput with low packet loss ratio can be achieved when the power of the received packets is balanced. Deviation from this condition leads to a rapid performance degradation. Balancing the packet power at the gateway demodulator input is not straightforward and, again, it requires some level of coordination, resulting in the aforementioned need to exchange signaling information. Otherwise, because of the fact that the gain of the satellite receiving antenna within its coverage area is not uniform and that packets sent by terminals located in different positions will undergo different levels of link attenuation caused by different propagation conditions, even if two terminals transmit with the same power, their packets will have different power levels when they reach the gateway demodulator.
A change of paradigm was therefore introduced in the design of the S-MIM uplink demodulator by exploiting the inherent presence of power unbalance at the demodulator input to enhance the system throughput. This is achieved through iSIC, which extends the conventional successive interference cancellation (SIC) technique, adapting it to the asynchronous and bursty nature of the transmitted packets. The demodulator stores in its memory a portion of the received signal, which is the superposition of many incoming packets, and, in the first iteration, attempts to detect and decode the packet with the highest received power. Then, this packet is cancelled from the received signal stored in the demodulator memory. During the second iteration, the packet with the next highest received power is detected and decoded. The process is repeated several times to resolve as many collisions as possible. Finally, the content of the demodulator memory is partly updated by storing a new portion of the received signal according to a sliding window approach and the process is repeated. As shown in Fig. 3, SSA combined with iSIC can allow, with truly affordable implementation complexity, a maximum throughput close to 1.8 bits/chip, about 3 dB below what is achievable with code-division multiple access (CDMA) with random spreading capacity. Further improvements can be obtained by optimizing the statistical distribution of the received power level at the demodulator, since it has been proved that a uniform distribution in decibels maximizes the performance of the iSIC algorithm, whereas the results in Fig. 3 correspond to a lognormal distribution. An open-loop power control algorithm can be used by all terminals to set their power so as to obtain an effective power distribution at the demodulator as close as possible to the optimum one.