Software Patent of the Week: Wireless + Dining Philosophers = Non-Obvious?

by Tim Lee on January 19, 2007 · Comments

Ordinarily, my software patent series focuses on patents that have been granted by the patent office and the subject of litigation. I’m going to break that pattern this week because reader Richard Bennett pointed to one of his own patent applications as an example of a worthwhile software patent. Since I frequently ask supporters of software patents to point out a good one (a request that’s almost always ignored) I thought I’d analyze Bennett’s patent application to see what we can learn. Below the cut are my thoughts on it.


Now, I’m at a little bit of an disadvantage in this case because the subject of the patent–low-level wireless networking–is not one I know well. That makes it difficult to judge the obviousness of the more technical aspects of the patent. However, it strikes me that you don’t have to know very much about wireless technology to see that it’s broader claims, at least, are too obvious to merit a patent.

Here is the patent application, and here is the first claim:

In a wireless network incorporating a medium access control (MAC) protocol, a MAC reservation-sensing mechanism for facilitating network use by a plurality of stations, comprising: listening to network traffic by a first station desiring to transmit on the network; determining the existence of reservation time slots on the network; identifying free time slots; broadcasting a reservation request for an identified free time slot to a second station on the network; coordinating, between the first station and the second station, to confirm that the identified free time slot is available; and iff there is no conflict on the network for the identified free time slot and the identified free time slot is confirmed to be free, transmitting data from the first station to the second station.

Subsequent claims describe a variety of additional strategies that a broadcasting station might use to deal with resource contention.

It’s not clear to me what’s novel or non-obvious about these claims. There’s a voluminous literature in computer science on concurrency and access to shared resources. For example, as an undergrad, I encountered the dining philosophers problem, which is a case study in coordinating among five agents (the philosophers) who must coordinate the use of five shared resources (forks) without a centralized decision maker. That’s obviously not quite the situation under discussion here, but it’s not too terribly wide of the mark. I’d bet money that someone who knows the relevant literature well could point to papers dealing with exactly the kind decentralized coordination problems Bennett’s patent is designed to solve.

Of course, it’s possible that none of those papers deal specifically with coordination in the context of MAC protocol wireless networks. But I don’t see how that renders the idea non-obvious. There’s nothing magical about wireless networking. The techniques described here to reserve wireless time slots would almost certainly be recognizable to people writing software to manage other kinds of shared resources in a computer system.

I should emphasize that I mean no disrespect to Mr. Bennett, who is, I’m sure a skilled network engineer who is developing some great products. But I just don’t see how giving Mr. Bennett a fairly broad monopoly on devices that negotiate access to a wireless network in a decentralized manner is consistent with the purposes of patent law. It will doubtless be good for Mr. Bennett and for the lawyers who filed and will litigate the patent, but I don’t see how it’s good for the rest of us.

Comments Posted in: Patents

  • You obviously don't understand how the IBM Token Ring works.

    It is not a slotted system. The Cambridge Ring, Sonet, T1, and ATM are slotted, but the packets in Token Ring are too big and the latency too low to allow slotting. One station takes the ring, and that station has exclusive access until the end of his packet. In a slotted system, accesses are of a fixed size and on a fixed schedule.

    "Early Token Release" is not a reservation system, it's a method of releasing the token at the end of the packet rather than waiting for the packet header to come back around to the originating station. It does allow a token to co-exist with a packet, but not with another token. There are no slots in this system, only packets.

    The Token Ring priority system (a means for bridges to take priority over normal stations) is a packet-by-packet system, not a periodic system like my invention.

    So none of the key elements of this system exist in Token Ring. The closest analog in wireline networking is something like the telephony networks where calls are allocated slots (or channels) in periodic packets. Those systems use a centralized form of control, not a negotiation system like my deal.

    In summary, you couldn't be farther from the mark.
  • Mark Seecof
    Um, pardon me whilst I, glumph, urkk, chew down the rest of my crow... Of course, Bennett's claim 1 specifically claims "a wireless network," so that would let 802.5 out--contrary to my flip assertion above. I apologize, Mr. Bennett. But leaving that qualifier aside (and it can be left aside for the purpose of assessing "novelty"), I stand by my analysis.

    I'm sure the next defense of Mr. Bennett's patent application will be the "inverted pyramid" of claim drafting. Of course, I will be told, claim 1 is ridiculously broad. The rest of the claims will zero in on the real invention. All I can say to that is, such an approach is bogus. If the leading claim is non-novel, then it shouldn't be there. Put the justifications and background material where they belong, not in the claims. Yes, that is a policy opinion rather than a view of the present state of the law.

  • Mark Seecof

    Mr. Bennett: It's not a strawman--you wrote that you considered a search for other patents sufficient. I thought you might have oversimplified your own views, so I wrote "seems to think." Perhaps you do hold more complex views, but you were the one who chose which portion of them to emphasize. (I think you are way too defensive. Why?)


    As for whether your claim 1 resembles Token Ring... Token Ring has time slots. It has more than one with early token release. It has broadcasting to reserve timeslots (the priority/reservation scheme), which also uses coordination between stations to resolve conflicts... (Your claim 1 doesn't have a "two-way handshake" either--it has "coordinating..." between stations. So does Token Ring.)


    I'll bet you a nickel that if you had a patent with claim 1 in force when I came out with an 802.5 LAN, you would assert that my product infringed your claim 1.


    Look, I said I hadn't read all your other claims. I have no doubt you built something rather different from Token Ring. But that doesn't mean (a) that it was really novel rather than an elaboration of existing technics, or (b) that it's an example of a "good" software patent.


    Gentle readers, I invite you to compare Bennett's claim 1 with 802.5 Token Ring. Decide for yourselves whether there is any resemblance.

  • Nice strawman, Mark. Nowhere do I claim that "prior art" is limited to patents; published papers and public presentations are also evidence of prior art, and you'll find many references to them in the patent database.

    Your assertion that my claim #1 is IBM Token Ring is ridiculous, frankly. I was one of the founders of the Open Token Foundation and a former Token Ring implementor, so I'm quite familiar with its method of operation.

    My claim #1 says: 1. In a wireless network incorporating a medium access control (MAC) protocol, a MAC reservation-sensing mechanism for facilitating network use by a plurality of stations, comprising: listening to network traffic by a first station desiring to transmit on the network; determining the existence of reservation time slots on the network; identifying free time slots; broadcasting a reservation request for an identified free time slot to a second station on the network; coordinating, between the first station and the second station, to confirm that the identified free time slot is available; and iff there is no conflict on the network for the identified free time slot and the identified free time slot is confirmed to be free, transmitting data from the first station to the second station.

    The only feature of this claim shared by Token Ring is listening before speaking. Token Ring doesn't have time slots, it doesn't have ex-ante reservations, it doesn't have a two-way handshake, and it doesn't use broadcasts in any way for medium access.

    So other than the 99% of the system that's not at all like Token Ring, you're right on - it has something to do with networks, for sure.
  • Mark Seecof

    Richard Bennett writes:

    I maintain that my solution is novel, in the context of the invention. I searched the database - as did a paid search company - for infringing patents on this solution and didn't find any.


    And that reveals part of the problem. Like the USPTO, Richard Bennett seems to think that only other patents constitute prior art. He also seems to think that anything which has not been patented is ipso facto novel.


    I suggest that both of those suppositions are wrong. If they're not wrong under the current precedents of the Federal Circuit, that just shows how many Federal Circuit cases were wrongly decided.


    As for Bennett's patent application, I have only read the portion excerpted here, but I see that the first claim appears to describe good old 802.5 Token Ring (see this brief description). I suppose it is a nice question whether some portions of Bennett's mechanisms are really not obvious to someone skilled in the art, but overall, his first claim looks pretty obvious to me.


    Note that I don't say the issue is that Bennett's problem is well known--patents are about solutions. I say Bennett's solution (his MAC scheme) to the problem (network contention) appears to be nothing more than good workmanship, elaborating and implementing general methods already well known in the industry.


    For different reasons, I agree with Mr. Lee: this patent application does not provide a clear example of a "good" software patent.

  • Right, the method of search automatically cherry-picks. We can be reasonably certain that the patents subject to litigation are the most questionable ones, that's the way the system is supposed to work. I think the more interesting question is about the mix of good software patents to bad ones, but that's a really hard question to answer.
  • Richard, I've been pretty much just typing "software patent" into Google News and writing about the first patent subject to litigation that I find. If there has been recent major software patent litigation that I haven't discussed, please let me know and I'll look at the patents involved in that case.
  • In other words, it's simply cherry-picking.
  • As is, Tim, you spend 97% of your patent of the week series explaining a patent, and only add a couple sentences talking about policy.

    ***Once I've done enough of them that I feel comfortable drawing generalizations***

    In 10 years, you will not have even covered 1% of software patents...
  • Tim says: In contrast, it seems to me that if this problem has been solved in some domain--say reserving space on a shared bus in a PC--that applying those same contexts to the wireless content would be obvious to someone of ordinary skill in the art.

    If the problems were the same in different domains, the a common solution would fail the obviousness test the second time it was applied. But the problems we face in wireless networking tend to differ in significant ways from the problems we find in nailed-up networking or computer buses. In wireless networking, messages transfers are generally much less reliable, power-saving is an issue (all those battery-powered devices) and we have to deal with mobility. These problems mandate variations and refinements to nailed-up networking solutions that push these inventions into novel territory.

    This scheme, by the way, was adopted by the MBOA and WiMedia Alliance for UWB networking, and you'll probably use it in your home in the next few years. There's not necessarily going to be a lot of royalty money changing hands, and if there is I won't see any of it.
  • Noel, you ask so many questions that it would take hours to answer them all. And in my experience, when I do answer your questions, you follow them up with a bunch more questions.

    In this particular case, most of your comment above doesn't actually ask any questions. To answer the one question I do see: the purpose of this series is to document the extent of the problem with the current patent system. Once I've done enough of them that I feel comfortable drawing generalizations, I might start trying to formulate a proposal for how things might be fixed.
  • Back to the basics Tim. I believe my questions above are fundemental to your patent of the week series. I wish you would simply address them.
  • Richard, you're right that the novelty requirement applies to the solution, rather than the problem. And it's certainly possible that no one has thought of solving a resource contention problem by having agents negotiate for future time slots in a decentralized manner. I don't know the relevant literature well enough to point to a specific paper describing this solution to the problem.

    However, where I think the current patent jurisprudence goes wrong is that current law would only recognize prior art if it were specific to wireless networking applications. In contrast, it seems to me that if this problem has been solved in some domain--say reserving space on a shared bus in a PC--that applying those same contexts to the wireless content would be obvious to someone of ordinary skill in the art.

    I do agree with Gary that this application has a more plausible case for novelty and non-obviousness than most of the ones I've looked at so far.
  • As I read it, it describes a protocol for collision detection and retrying without a central arbiter. In spite of the claim that a "centralized reservation station [is] required by current practice," this isn't a novel idea in itself. The novelty appears to be that it tries to anticipate future time slot allocation and reserve a slot in advance.

    I'm not up on research in protocols, but this does sound like a novel and clever idea to me if it can be made to work. I've come to object to software patents on broader grounds, but this looks more reasonable than most.
  • The problem of decentralized decision-making is certainly not novel, but I maintain that my solution is novel, in the context of the invention. I searched the database - as did a paid search company - for infringing patents on this solution and didn't find any.

    So once again, "novelty" isn't an attribute of the problem, it applies to the solution. And if you think the solution isn't novel, patent law says it's not sufficient for you to have a feeling, you need to support it with evidence.

    This being a real-world example, you should have no problem doing so, if you're right.
  • Oh, I'm sorry, you wrote that you "don't see" how Richard's patent is valuable. Again, you quibble about the small stuff knowing full well my point.
  • Noel, in the English language, the quote symbol is used to set off exact quotations--i.e. things that the person you're quoting actually said and wrote. Now, unless the "search" function on my browser is broken, the only person who's used the phrase "I don't understand" on this page is you. So what on earth are you talking about?
  • Oh yeah, the phrase "I don't understand" as you use above, about Richard Bennet's patent needs to be expanded on. When someone who is knowledgeable says they "don't understand" then clarification is due. But when someone who is not knowledgeable says they "don't understand," well...
  • Tim, again, explain your standards for novelty and obviousness, besides just saying how you don't see any point in the USPTO giving Richard Bennet this patent.

    I don't think you understand the point of the patent system.

    First, you only consider the ex ante justification (if you don't know what that is, I'm not going to bother telling you).

    Second, you seem to think that each patent itself represents an innovation; thats not the case.

    Third, what is the point of this patent of the week series when you don't tie your analysis to policy.

    Forth, if you think all of these patents are obvious and not novel, then why are you not out there commercializing all the concepts they encompass.
blog comments powered by Disqus

Previous post:

Next post: